Is a distribution that is normal, but highly skewed, considered Gaussian? The 2019 Stack Overflow Developer Survey Results Are In Announcing the arrival of Valued Associate #679: Cesar Manara Planned maintenance scheduled April 17/18, 2019 at 00:00UTC (8:00pm US/Eastern)Is normality testing 'essentially useless'?What is the difference between zero-inflated and hurdle models?If my histogram shows a bell-shaped curve, can I say my data is normally distributed?What would the distribution of time spent per day on a given site look like?How do I identify the “Long Tail” portion of my distribution?Skewed but bell-shaped still considered as normal distribution for ANOVA?Which Distribution Does the Data Point Belong to?Skewness - why is this distribution right skewed?log transform vs. resamplingIs it valid to remove the overhead of finding the current time for a computer program this way?Histograms for severely skewed dataWhat would the distribution of time spent per day on a given site look like?Distinguish between underlying Distribution and data shape in data transforming?Using bootstrap to estimate the 95th percentile and confidence interval for skewed data
How do you keep chess fun when your opponent constantly beats you?
Can I visit the Trinity College (Cambridge) library and see some of their rare books
Do warforged have souls?
different output for groups and groups USERNAME after adding a username to a group
How to read αἱμύλιος or when to aspirate
Free operad over a monoid object
Simulating Exploding Dice
How to support a colleague who finds meetings extremely tiring?
Can withdrawing asylum be illegal?
Is it ethical to upload a automatically generated paper to a non peer-reviewed site as part of a larger research?
Is an up-to-date browser secure on an out-of-date OS?
What can I do if neighbor is blocking my solar panels intentionally?
What does Linus Torvalds mean when he says that Git "never ever" tracks a file?
Working through Single Responsibility Principle in Python when Calls are Expensive
What other Star Trek series did the main TNG cast show up in?
Categorical vs continuous feature selection/engineering
What happens to a Warlock's expended Spell Slots when they gain a Level?
Why did Peik Lin say, "I'm not an animal"?
How do spell lists change if the party levels up without taking a long rest?
Solving overdetermined system by QR decomposition
Can each chord in a progression create its own key?
Nested ellipses in tikzpicture: Chomsky hierarchy
Did the new image of black hole confirm the general theory of relativity?
Match Roman Numerals
Is a distribution that is normal, but highly skewed, considered Gaussian?
The 2019 Stack Overflow Developer Survey Results Are In
Announcing the arrival of Valued Associate #679: Cesar Manara
Planned maintenance scheduled April 17/18, 2019 at 00:00UTC (8:00pm US/Eastern)Is normality testing 'essentially useless'?What is the difference between zero-inflated and hurdle models?If my histogram shows a bell-shaped curve, can I say my data is normally distributed?What would the distribution of time spent per day on a given site look like?How do I identify the “Long Tail” portion of my distribution?Skewed but bell-shaped still considered as normal distribution for ANOVA?Which Distribution Does the Data Point Belong to?Skewness - why is this distribution right skewed?log transform vs. resamplingIs it valid to remove the overhead of finding the current time for a computer program this way?Histograms for severely skewed dataWhat would the distribution of time spent per day on a given site look like?Distinguish between underlying Distribution and data shape in data transforming?Using bootstrap to estimate the 95th percentile and confidence interval for skewed data
.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty margin-bottom:0;
$begingroup$
I have this question: What do you think the distribution of time spent per day on YouTube looks like?
My answer is that it is probably normally distributed and highly left skewed. I expect there is one mode where most users spend around some average time and then a long right tail since some users are overwhelming power users.
Is that a fair answer? Is there a better word for that distribution?
distributions normal-distribution skewness skew-normal
$endgroup$
|
show 1 more comment
$begingroup$
I have this question: What do you think the distribution of time spent per day on YouTube looks like?
My answer is that it is probably normally distributed and highly left skewed. I expect there is one mode where most users spend around some average time and then a long right tail since some users are overwhelming power users.
Is that a fair answer? Is there a better word for that distribution?
distributions normal-distribution skewness skew-normal
$endgroup$
4
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
1
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
5
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
2
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53
|
show 1 more comment
$begingroup$
I have this question: What do you think the distribution of time spent per day on YouTube looks like?
My answer is that it is probably normally distributed and highly left skewed. I expect there is one mode where most users spend around some average time and then a long right tail since some users are overwhelming power users.
Is that a fair answer? Is there a better word for that distribution?
distributions normal-distribution skewness skew-normal
$endgroup$
I have this question: What do you think the distribution of time spent per day on YouTube looks like?
My answer is that it is probably normally distributed and highly left skewed. I expect there is one mode where most users spend around some average time and then a long right tail since some users are overwhelming power users.
Is that a fair answer? Is there a better word for that distribution?
distributions normal-distribution skewness skew-normal
distributions normal-distribution skewness skew-normal
edited Mar 31 at 6:46
Nick Cox
39.3k587131
39.3k587131
asked Mar 30 at 19:14
CauderCauder
10318
10318
4
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
1
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
5
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
2
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53
|
show 1 more comment
4
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
1
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
5
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
2
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53
4
4
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
1
1
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
5
5
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
2
2
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53
|
show 1 more comment
11 Answers
11
active
oldest
votes
$begingroup$
A fraction per day is certainly not negative. This rules out the normal distribution, which has probability mass over the entire real axis - in particular over the negative half.
Power law distributions are often used to model things like income distributions, sizes of cities etc. They are nonnegative and typically highly skewed. These would be the first I would try in modeling time spent watching YouTube. (Or monitoring CrossValidated questions.)
More information on power laws can be found here or here, or in our power-law tag.
$endgroup$
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
add a comment |
$begingroup$
A distribution that is normal is not highly skewed. That is a contradiction. Normally distributed variables have skew = 0.
$endgroup$
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
add a comment |
$begingroup$
If it has long right tail, then it's right skewed.
It can't be a normal distribution since skew !=0, it's perhaps a unimodal skew normal distribution:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
add a comment |
$begingroup$
It could be a log-normal distribution. As mentioned here:
Users' dwell time on online articles (jokes, news etc.) follows a log-normal distribution.
The reference given is: Yin, Peifeng; Luo, Ping; Lee, Wang-Chien; Wang, Min (2013). Silence is also evidence: interpreting dwell time for recommendation from psychological perspective. ACM International Conference on KDD.
$endgroup$
add a comment |
$begingroup$
The gamma distribution could be a good candidate to describe this kind of distribution over nonnegative, right-skewed data. See the green line in the image here:
https://en.m.wikipedia.org/wiki/Gamma_distribution
$endgroup$
add a comment |
$begingroup$
"Is there a better word for that distribution?"
There's a worthwhile distinction here between using words to describe the properties of the distribution, versus trying to find a "name" for the distribution so that you can identify it as (approximately) an instance of a particular standard distribution: one for which a formula or statistical tables might exist for its distribution function, and for which you could estimate its parameters. In this latter case, you are likely using the named distribution, e.g. "normal/Gaussian" (the two terms are generally synonymous), as a model that captures some of the key features of your data, rather than claiming the population your data is drawn from exactly follows that theoretical distribution. To slightly misquote George Box, all models are "wrong", but some are useful. If you are thinking about the modelling approach, it is worth considering what features you want to incorporate and how complicated or parsimonious you want your model to be.
Being positively skewed is an example of describing a property that the distribution has, but doesn't come close to specifying which off-the-shelf distribution is "the" appropriate model. It does rule out some candidates, for example the Gaussian (i.e. normal) distribution has zero skew so will not be appropriate to model your data if the skew is an important feature. There may be other properties of the data that are important to you too, e.g. that it's unimodal (has just one peak) or that it is bounded between 0 and 24 hours (or between 0 and 1, if you are writing it as a fraction of the day), or that there is a probability mass concentrated at zero (since there are people who do not watch youtube at all on a given day). You may also be interested in other properties like the kurtosis. And it is worth bearing in mind that even if your distribution had a "hump" or "bell-curve" shape and had zero or near-zero skew, it doesn't automatically follow that the normal distribution is "correct" for it! On the other hand, even if the population your data is drawn from actually did follow a particular distribution precisely, due to sampling error your dataset may not quite resemble it. Small data sets are likely to be "noisy", and it may be unclear whether certain features you can see, e.g. additional small humps or asymmetric tails, are properties of the underlying population the data was drawn from (and perhaps therefore ought to be incorporated in your model) or whether they are just artefacts from your particular sample (and for modelling purposes should be ignored). If you have a small data set and the skew is close to zero, then it is even plausible the underlying distribution is actually symmetric. The larger your data set and the larger the skewness, the less plausible this becomes — but while you could perform a significance test to see how convincing is the evidence your data provides for skewness in the population it was drawn from, this may be missing the point as to whether a normal (or other zero skew) distribution is appropriate as a model ...
Which properties of the data really matter for the purposes you are intending to model it? Note that if the skew is reasonably small and you do not care very much about it, even if the underlying population is genuinely skewed, then you might still find the normal distribution a useful model to approximate this true distribution of watching times. But you should check that this doesn't end up making silly predictions. Because a normal distribution has no highest or lowest possible value, then although extremely high or low values become increasingly unlikely, you will always find that your model predicts there is some probability of watching for a negative number of hours per day, or more than 24 hours. This gets more problematic for you if the predicted probability of such impossible events becomes high. A symmetric distribution like the normal will predict that as many people will watch for lengths of time more than e.g. 50% above the mean, as watch for less than 50% below the mean. If watching times are very skewed, then this kind of prediction may also be so implausible as to be silly, and give you misleading results if you are taking the results of your model and using them as inputs for some other purpose (for instance, you're running a simulation of watching times in order to calculate optimal advertisement scheduling). If the skewness is so noteworthy you want to capture it as part of your model, then the skew normal distribution may be more appropriate. If you want to capture both skewness and kurtosis, then consider the skewed t. If you want to incorporate the physically possible upper and lower bounds, then consider using the truncated versions of these distributions. Many other probability distributions exist that can be skewed and unimodal (for appropriate parameter choices) such as the F or gamma distributions, and again you can truncate these so they do not predict impossibly high watching times. A beta distribution may be a good choice if you are modelling the fraction of the day spent watching, as this is always bounded between 0 and 1 without further truncation being necessary. If you want to incorporate the concentration of probability at exactly zero due to non-watchers, then consider building in a hurdle model.
But at the point you are trying to throw in every feature you can identify from your data, and build an ever more sophisticated model, perhaps you should ask yourself why you are doing this? Would there be an advantage to a simpler model, for example it being easier to work with mathematically or having fewer parameters to estimate? If you are concerned that such simplification will leave you unable to capture all of the properties of interest to you, it may well be that no "off-the-shelf" distribution does quite what you want. However, we are not restricted to working with named distributions whose mathematical properties have been elucidated previously. Instead, consider using your data to construct an empirical distribution function. This will capture all the behaviour that was present in your data, but you can no longer give it a name like "normal" or "gamma", nor can you apply mathematical properties that pertain only to a particular distribution. For instance, the "95% of the data lies within 1.96 standard deviations of the mean" rule is for normally distributed data and may not apply to your distribution; though note that some rules apply to all distributions, e.g. Chebyshev's inequality guarantees at least 75% of your data must lie within two standard deviations of the mean, regardless of the skew. Unfortunately the empirical distribution will also inherit all those properties of your data set arising purely by sampling error, not just those possessed by the underlying population, so you may find a histogram of your empirical distribution has some humps and dips that the population itself does not. You may want to investigate smoothed empirical distribution functions, or better yet, increasing your sample size.
In summary: although the normal distribution has zero skew, the fact your data are skewed doesn't rule out the normal distribution as a useful model, though it does suggest some other distribution may be more appropriate. You should consider other properties of the data when choosing your model, besides the skew, and consider too the purposes you are going to use the model for. It's safe to say that your true population of watching times does not exactly follow some famous, named distribution, but this does not mean such a distribution is doomed to be useless as a model. However, for some purposes you may prefer to just use the empirical distribution itself, rather than try fitting a standard distribution to it.
$endgroup$
add a comment |
$begingroup$
"Normal" and "Gaussian" mean exactly the same thing. As other answers explain, the distribution you're talking about is not normal/Gaussian, because that distribution assigns probabilities to every value on the real line, whereas your distribution only exists between $0$ and $24$.
$endgroup$
add a comment |
$begingroup$
In the case at hand, since the time spent per day is bound from $0$ to $1$ (if quantified as a fraction of the day), distributions that are unbounded above (e.g. Pareto, skew-normal, Gamma, log-normal) won't work, but Beta would.
$endgroup$
add a comment |
$begingroup$
How about a hurdle model?
A hurdle model has two parts. The first is Bernoulli experiment that determines whether you use YouTube at all. If you don't, then your usage time is obviously zero and you're done. If you do, you "pass that hurdle", then the usage time comes from some other strictly positive distribution.
A closely related concept are zero-inflated models. These are meant to deal with a situation where we observe a bunch of zeros, but can't distinguish between always-zeros and sometimes-zeros. For example, consider the number of cigarettes that a person smokes each day. For non-smokers, that number is always zero, but some smokers may not smoke on a given day (out of cigarettes? on a long flight?). Unlike the hurdle model, the "smoker" distribution here should include zero, but these counts are 'inflated' by the non-smokers' contribution too.
$endgroup$
add a comment |
$begingroup$
If the distribution is indeed a 'subset' of the normal distribution, you should considder a truncated model. Widely used in this context is the family of TOBIT models.
They essentialy suggest a pdf with a (positive) probability mass at 0 and then a 'cut of part of the normal distribution' for positive values.
I will refrain from typing the formula here and rather refere you to the Wikipedia Article: https://en.wikipedia.org/wiki/Tobit_model
$endgroup$
add a comment |
$begingroup$
Normal distributions are by definition non-skewed, so you can't have both things. If the distribution is left-skewed, then it cannot be Gaussian. You'll have to pick a different one! The closest thing to your request I can think of is this:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
add a comment |
protected by gung♦ Apr 2 at 13:16
Thank you for your interest in this question.
Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).
Would you like to answer one of these unanswered questions instead?
11 Answers
11
active
oldest
votes
11 Answers
11
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
A fraction per day is certainly not negative. This rules out the normal distribution, which has probability mass over the entire real axis - in particular over the negative half.
Power law distributions are often used to model things like income distributions, sizes of cities etc. They are nonnegative and typically highly skewed. These would be the first I would try in modeling time spent watching YouTube. (Or monitoring CrossValidated questions.)
More information on power laws can be found here or here, or in our power-law tag.
$endgroup$
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
add a comment |
$begingroup$
A fraction per day is certainly not negative. This rules out the normal distribution, which has probability mass over the entire real axis - in particular over the negative half.
Power law distributions are often used to model things like income distributions, sizes of cities etc. They are nonnegative and typically highly skewed. These would be the first I would try in modeling time spent watching YouTube. (Or monitoring CrossValidated questions.)
More information on power laws can be found here or here, or in our power-law tag.
$endgroup$
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
add a comment |
$begingroup$
A fraction per day is certainly not negative. This rules out the normal distribution, which has probability mass over the entire real axis - in particular over the negative half.
Power law distributions are often used to model things like income distributions, sizes of cities etc. They are nonnegative and typically highly skewed. These would be the first I would try in modeling time spent watching YouTube. (Or monitoring CrossValidated questions.)
More information on power laws can be found here or here, or in our power-law tag.
$endgroup$
A fraction per day is certainly not negative. This rules out the normal distribution, which has probability mass over the entire real axis - in particular over the negative half.
Power law distributions are often used to model things like income distributions, sizes of cities etc. They are nonnegative and typically highly skewed. These would be the first I would try in modeling time spent watching YouTube. (Or monitoring CrossValidated questions.)
More information on power laws can be found here or here, or in our power-law tag.
answered Mar 30 at 19:35
Stephan KolassaStephan Kolassa
47.5k7101178
47.5k7101178
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
add a comment |
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
14
14
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
$begingroup$
You're completely correct that normal distributions have support on the real line. And yet...they're no an awful model for some strictly positive qualities, like adults' height or weight, where the mean and variance are such that the negative values are very unlikely under the model.
$endgroup$
– Matt Krause
Mar 30 at 22:26
2
2
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause That's actually a great question - is there a same probability I will be '10 cm above or below the mean height' or '10 percent above or below the mean height'? Only the first case could warrant normal distribution.
$endgroup$
– Tomáš Kafka
Apr 1 at 12:26
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
$begingroup$
@MattKrause: I completely agree, in a general sense. Yet, the present question is about the proportion of daily time spent watching YouTube. We don't have any data, but I would be extremely surprised if the distribution was even remotely symmetric.
$endgroup$
– Stephan Kolassa
Apr 1 at 15:28
add a comment |
$begingroup$
A distribution that is normal is not highly skewed. That is a contradiction. Normally distributed variables have skew = 0.
$endgroup$
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
add a comment |
$begingroup$
A distribution that is normal is not highly skewed. That is a contradiction. Normally distributed variables have skew = 0.
$endgroup$
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
add a comment |
$begingroup$
A distribution that is normal is not highly skewed. That is a contradiction. Normally distributed variables have skew = 0.
$endgroup$
A distribution that is normal is not highly skewed. That is a contradiction. Normally distributed variables have skew = 0.
answered Mar 30 at 19:15
Peter Flom♦Peter Flom
77.3k12109217
77.3k12109217
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
add a comment |
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
1
1
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
$begingroup$
What is a better way to describe the distribution? Is there a word for that type of distribution where it centers around a mode and then has a long tail?
$endgroup$
– Cauder
Mar 30 at 19:21
13
13
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
$begingroup$
Unimodal and skewed is as close as I can come...
$endgroup$
– jbowman
Mar 30 at 19:27
9
9
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
$begingroup$
As an aside, it's just really incredible that people give their time to help other people get better at this stuff. I know it goes without saying, but it's so cool what you both do!
$endgroup$
– Cauder
Mar 30 at 19:30
6
6
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
Yes, but it's worth clarifying that that statement pertains to the normally distributed population. A sample drawn from that population can be very skewed.
$endgroup$
– gung♦
Mar 31 at 2:14
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
$begingroup$
When the skew value is small ("small" being decided by the people dealing with the stats in question), you can still treat the population as normal, albeit with minor error as a result.
$endgroup$
– Carl Witthoft
Apr 1 at 18:03
add a comment |
$begingroup$
If it has long right tail, then it's right skewed.
It can't be a normal distribution since skew !=0, it's perhaps a unimodal skew normal distribution:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
add a comment |
$begingroup$
If it has long right tail, then it's right skewed.
It can't be a normal distribution since skew !=0, it's perhaps a unimodal skew normal distribution:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
add a comment |
$begingroup$
If it has long right tail, then it's right skewed.
It can't be a normal distribution since skew !=0, it's perhaps a unimodal skew normal distribution:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
If it has long right tail, then it's right skewed.
It can't be a normal distribution since skew !=0, it's perhaps a unimodal skew normal distribution:
https://en.wikipedia.org/wiki/Skew_normal_distribution
answered Mar 30 at 19:31
beholdbehold
3599
3599
add a comment |
add a comment |
$begingroup$
It could be a log-normal distribution. As mentioned here:
Users' dwell time on online articles (jokes, news etc.) follows a log-normal distribution.
The reference given is: Yin, Peifeng; Luo, Ping; Lee, Wang-Chien; Wang, Min (2013). Silence is also evidence: interpreting dwell time for recommendation from psychological perspective. ACM International Conference on KDD.
$endgroup$
add a comment |
$begingroup$
It could be a log-normal distribution. As mentioned here:
Users' dwell time on online articles (jokes, news etc.) follows a log-normal distribution.
The reference given is: Yin, Peifeng; Luo, Ping; Lee, Wang-Chien; Wang, Min (2013). Silence is also evidence: interpreting dwell time for recommendation from psychological perspective. ACM International Conference on KDD.
$endgroup$
add a comment |
$begingroup$
It could be a log-normal distribution. As mentioned here:
Users' dwell time on online articles (jokes, news etc.) follows a log-normal distribution.
The reference given is: Yin, Peifeng; Luo, Ping; Lee, Wang-Chien; Wang, Min (2013). Silence is also evidence: interpreting dwell time for recommendation from psychological perspective. ACM International Conference on KDD.
$endgroup$
It could be a log-normal distribution. As mentioned here:
Users' dwell time on online articles (jokes, news etc.) follows a log-normal distribution.
The reference given is: Yin, Peifeng; Luo, Ping; Lee, Wang-Chien; Wang, Min (2013). Silence is also evidence: interpreting dwell time for recommendation from psychological perspective. ACM International Conference on KDD.
answered Mar 31 at 1:01
Count IblisCount Iblis
28114
28114
add a comment |
add a comment |
$begingroup$
The gamma distribution could be a good candidate to describe this kind of distribution over nonnegative, right-skewed data. See the green line in the image here:
https://en.m.wikipedia.org/wiki/Gamma_distribution
$endgroup$
add a comment |
$begingroup$
The gamma distribution could be a good candidate to describe this kind of distribution over nonnegative, right-skewed data. See the green line in the image here:
https://en.m.wikipedia.org/wiki/Gamma_distribution
$endgroup$
add a comment |
$begingroup$
The gamma distribution could be a good candidate to describe this kind of distribution over nonnegative, right-skewed data. See the green line in the image here:
https://en.m.wikipedia.org/wiki/Gamma_distribution
$endgroup$
The gamma distribution could be a good candidate to describe this kind of distribution over nonnegative, right-skewed data. See the green line in the image here:
https://en.m.wikipedia.org/wiki/Gamma_distribution
answered Mar 31 at 6:00
mauricemaurice
19816
19816
add a comment |
add a comment |
$begingroup$
"Is there a better word for that distribution?"
There's a worthwhile distinction here between using words to describe the properties of the distribution, versus trying to find a "name" for the distribution so that you can identify it as (approximately) an instance of a particular standard distribution: one for which a formula or statistical tables might exist for its distribution function, and for which you could estimate its parameters. In this latter case, you are likely using the named distribution, e.g. "normal/Gaussian" (the two terms are generally synonymous), as a model that captures some of the key features of your data, rather than claiming the population your data is drawn from exactly follows that theoretical distribution. To slightly misquote George Box, all models are "wrong", but some are useful. If you are thinking about the modelling approach, it is worth considering what features you want to incorporate and how complicated or parsimonious you want your model to be.
Being positively skewed is an example of describing a property that the distribution has, but doesn't come close to specifying which off-the-shelf distribution is "the" appropriate model. It does rule out some candidates, for example the Gaussian (i.e. normal) distribution has zero skew so will not be appropriate to model your data if the skew is an important feature. There may be other properties of the data that are important to you too, e.g. that it's unimodal (has just one peak) or that it is bounded between 0 and 24 hours (or between 0 and 1, if you are writing it as a fraction of the day), or that there is a probability mass concentrated at zero (since there are people who do not watch youtube at all on a given day). You may also be interested in other properties like the kurtosis. And it is worth bearing in mind that even if your distribution had a "hump" or "bell-curve" shape and had zero or near-zero skew, it doesn't automatically follow that the normal distribution is "correct" for it! On the other hand, even if the population your data is drawn from actually did follow a particular distribution precisely, due to sampling error your dataset may not quite resemble it. Small data sets are likely to be "noisy", and it may be unclear whether certain features you can see, e.g. additional small humps or asymmetric tails, are properties of the underlying population the data was drawn from (and perhaps therefore ought to be incorporated in your model) or whether they are just artefacts from your particular sample (and for modelling purposes should be ignored). If you have a small data set and the skew is close to zero, then it is even plausible the underlying distribution is actually symmetric. The larger your data set and the larger the skewness, the less plausible this becomes — but while you could perform a significance test to see how convincing is the evidence your data provides for skewness in the population it was drawn from, this may be missing the point as to whether a normal (or other zero skew) distribution is appropriate as a model ...
Which properties of the data really matter for the purposes you are intending to model it? Note that if the skew is reasonably small and you do not care very much about it, even if the underlying population is genuinely skewed, then you might still find the normal distribution a useful model to approximate this true distribution of watching times. But you should check that this doesn't end up making silly predictions. Because a normal distribution has no highest or lowest possible value, then although extremely high or low values become increasingly unlikely, you will always find that your model predicts there is some probability of watching for a negative number of hours per day, or more than 24 hours. This gets more problematic for you if the predicted probability of such impossible events becomes high. A symmetric distribution like the normal will predict that as many people will watch for lengths of time more than e.g. 50% above the mean, as watch for less than 50% below the mean. If watching times are very skewed, then this kind of prediction may also be so implausible as to be silly, and give you misleading results if you are taking the results of your model and using them as inputs for some other purpose (for instance, you're running a simulation of watching times in order to calculate optimal advertisement scheduling). If the skewness is so noteworthy you want to capture it as part of your model, then the skew normal distribution may be more appropriate. If you want to capture both skewness and kurtosis, then consider the skewed t. If you want to incorporate the physically possible upper and lower bounds, then consider using the truncated versions of these distributions. Many other probability distributions exist that can be skewed and unimodal (for appropriate parameter choices) such as the F or gamma distributions, and again you can truncate these so they do not predict impossibly high watching times. A beta distribution may be a good choice if you are modelling the fraction of the day spent watching, as this is always bounded between 0 and 1 without further truncation being necessary. If you want to incorporate the concentration of probability at exactly zero due to non-watchers, then consider building in a hurdle model.
But at the point you are trying to throw in every feature you can identify from your data, and build an ever more sophisticated model, perhaps you should ask yourself why you are doing this? Would there be an advantage to a simpler model, for example it being easier to work with mathematically or having fewer parameters to estimate? If you are concerned that such simplification will leave you unable to capture all of the properties of interest to you, it may well be that no "off-the-shelf" distribution does quite what you want. However, we are not restricted to working with named distributions whose mathematical properties have been elucidated previously. Instead, consider using your data to construct an empirical distribution function. This will capture all the behaviour that was present in your data, but you can no longer give it a name like "normal" or "gamma", nor can you apply mathematical properties that pertain only to a particular distribution. For instance, the "95% of the data lies within 1.96 standard deviations of the mean" rule is for normally distributed data and may not apply to your distribution; though note that some rules apply to all distributions, e.g. Chebyshev's inequality guarantees at least 75% of your data must lie within two standard deviations of the mean, regardless of the skew. Unfortunately the empirical distribution will also inherit all those properties of your data set arising purely by sampling error, not just those possessed by the underlying population, so you may find a histogram of your empirical distribution has some humps and dips that the population itself does not. You may want to investigate smoothed empirical distribution functions, or better yet, increasing your sample size.
In summary: although the normal distribution has zero skew, the fact your data are skewed doesn't rule out the normal distribution as a useful model, though it does suggest some other distribution may be more appropriate. You should consider other properties of the data when choosing your model, besides the skew, and consider too the purposes you are going to use the model for. It's safe to say that your true population of watching times does not exactly follow some famous, named distribution, but this does not mean such a distribution is doomed to be useless as a model. However, for some purposes you may prefer to just use the empirical distribution itself, rather than try fitting a standard distribution to it.
$endgroup$
add a comment |
$begingroup$
"Is there a better word for that distribution?"
There's a worthwhile distinction here between using words to describe the properties of the distribution, versus trying to find a "name" for the distribution so that you can identify it as (approximately) an instance of a particular standard distribution: one for which a formula or statistical tables might exist for its distribution function, and for which you could estimate its parameters. In this latter case, you are likely using the named distribution, e.g. "normal/Gaussian" (the two terms are generally synonymous), as a model that captures some of the key features of your data, rather than claiming the population your data is drawn from exactly follows that theoretical distribution. To slightly misquote George Box, all models are "wrong", but some are useful. If you are thinking about the modelling approach, it is worth considering what features you want to incorporate and how complicated or parsimonious you want your model to be.
Being positively skewed is an example of describing a property that the distribution has, but doesn't come close to specifying which off-the-shelf distribution is "the" appropriate model. It does rule out some candidates, for example the Gaussian (i.e. normal) distribution has zero skew so will not be appropriate to model your data if the skew is an important feature. There may be other properties of the data that are important to you too, e.g. that it's unimodal (has just one peak) or that it is bounded between 0 and 24 hours (or between 0 and 1, if you are writing it as a fraction of the day), or that there is a probability mass concentrated at zero (since there are people who do not watch youtube at all on a given day). You may also be interested in other properties like the kurtosis. And it is worth bearing in mind that even if your distribution had a "hump" or "bell-curve" shape and had zero or near-zero skew, it doesn't automatically follow that the normal distribution is "correct" for it! On the other hand, even if the population your data is drawn from actually did follow a particular distribution precisely, due to sampling error your dataset may not quite resemble it. Small data sets are likely to be "noisy", and it may be unclear whether certain features you can see, e.g. additional small humps or asymmetric tails, are properties of the underlying population the data was drawn from (and perhaps therefore ought to be incorporated in your model) or whether they are just artefacts from your particular sample (and for modelling purposes should be ignored). If you have a small data set and the skew is close to zero, then it is even plausible the underlying distribution is actually symmetric. The larger your data set and the larger the skewness, the less plausible this becomes — but while you could perform a significance test to see how convincing is the evidence your data provides for skewness in the population it was drawn from, this may be missing the point as to whether a normal (or other zero skew) distribution is appropriate as a model ...
Which properties of the data really matter for the purposes you are intending to model it? Note that if the skew is reasonably small and you do not care very much about it, even if the underlying population is genuinely skewed, then you might still find the normal distribution a useful model to approximate this true distribution of watching times. But you should check that this doesn't end up making silly predictions. Because a normal distribution has no highest or lowest possible value, then although extremely high or low values become increasingly unlikely, you will always find that your model predicts there is some probability of watching for a negative number of hours per day, or more than 24 hours. This gets more problematic for you if the predicted probability of such impossible events becomes high. A symmetric distribution like the normal will predict that as many people will watch for lengths of time more than e.g. 50% above the mean, as watch for less than 50% below the mean. If watching times are very skewed, then this kind of prediction may also be so implausible as to be silly, and give you misleading results if you are taking the results of your model and using them as inputs for some other purpose (for instance, you're running a simulation of watching times in order to calculate optimal advertisement scheduling). If the skewness is so noteworthy you want to capture it as part of your model, then the skew normal distribution may be more appropriate. If you want to capture both skewness and kurtosis, then consider the skewed t. If you want to incorporate the physically possible upper and lower bounds, then consider using the truncated versions of these distributions. Many other probability distributions exist that can be skewed and unimodal (for appropriate parameter choices) such as the F or gamma distributions, and again you can truncate these so they do not predict impossibly high watching times. A beta distribution may be a good choice if you are modelling the fraction of the day spent watching, as this is always bounded between 0 and 1 without further truncation being necessary. If you want to incorporate the concentration of probability at exactly zero due to non-watchers, then consider building in a hurdle model.
But at the point you are trying to throw in every feature you can identify from your data, and build an ever more sophisticated model, perhaps you should ask yourself why you are doing this? Would there be an advantage to a simpler model, for example it being easier to work with mathematically or having fewer parameters to estimate? If you are concerned that such simplification will leave you unable to capture all of the properties of interest to you, it may well be that no "off-the-shelf" distribution does quite what you want. However, we are not restricted to working with named distributions whose mathematical properties have been elucidated previously. Instead, consider using your data to construct an empirical distribution function. This will capture all the behaviour that was present in your data, but you can no longer give it a name like "normal" or "gamma", nor can you apply mathematical properties that pertain only to a particular distribution. For instance, the "95% of the data lies within 1.96 standard deviations of the mean" rule is for normally distributed data and may not apply to your distribution; though note that some rules apply to all distributions, e.g. Chebyshev's inequality guarantees at least 75% of your data must lie within two standard deviations of the mean, regardless of the skew. Unfortunately the empirical distribution will also inherit all those properties of your data set arising purely by sampling error, not just those possessed by the underlying population, so you may find a histogram of your empirical distribution has some humps and dips that the population itself does not. You may want to investigate smoothed empirical distribution functions, or better yet, increasing your sample size.
In summary: although the normal distribution has zero skew, the fact your data are skewed doesn't rule out the normal distribution as a useful model, though it does suggest some other distribution may be more appropriate. You should consider other properties of the data when choosing your model, besides the skew, and consider too the purposes you are going to use the model for. It's safe to say that your true population of watching times does not exactly follow some famous, named distribution, but this does not mean such a distribution is doomed to be useless as a model. However, for some purposes you may prefer to just use the empirical distribution itself, rather than try fitting a standard distribution to it.
$endgroup$
add a comment |
$begingroup$
"Is there a better word for that distribution?"
There's a worthwhile distinction here between using words to describe the properties of the distribution, versus trying to find a "name" for the distribution so that you can identify it as (approximately) an instance of a particular standard distribution: one for which a formula or statistical tables might exist for its distribution function, and for which you could estimate its parameters. In this latter case, you are likely using the named distribution, e.g. "normal/Gaussian" (the two terms are generally synonymous), as a model that captures some of the key features of your data, rather than claiming the population your data is drawn from exactly follows that theoretical distribution. To slightly misquote George Box, all models are "wrong", but some are useful. If you are thinking about the modelling approach, it is worth considering what features you want to incorporate and how complicated or parsimonious you want your model to be.
Being positively skewed is an example of describing a property that the distribution has, but doesn't come close to specifying which off-the-shelf distribution is "the" appropriate model. It does rule out some candidates, for example the Gaussian (i.e. normal) distribution has zero skew so will not be appropriate to model your data if the skew is an important feature. There may be other properties of the data that are important to you too, e.g. that it's unimodal (has just one peak) or that it is bounded between 0 and 24 hours (or between 0 and 1, if you are writing it as a fraction of the day), or that there is a probability mass concentrated at zero (since there are people who do not watch youtube at all on a given day). You may also be interested in other properties like the kurtosis. And it is worth bearing in mind that even if your distribution had a "hump" or "bell-curve" shape and had zero or near-zero skew, it doesn't automatically follow that the normal distribution is "correct" for it! On the other hand, even if the population your data is drawn from actually did follow a particular distribution precisely, due to sampling error your dataset may not quite resemble it. Small data sets are likely to be "noisy", and it may be unclear whether certain features you can see, e.g. additional small humps or asymmetric tails, are properties of the underlying population the data was drawn from (and perhaps therefore ought to be incorporated in your model) or whether they are just artefacts from your particular sample (and for modelling purposes should be ignored). If you have a small data set and the skew is close to zero, then it is even plausible the underlying distribution is actually symmetric. The larger your data set and the larger the skewness, the less plausible this becomes — but while you could perform a significance test to see how convincing is the evidence your data provides for skewness in the population it was drawn from, this may be missing the point as to whether a normal (or other zero skew) distribution is appropriate as a model ...
Which properties of the data really matter for the purposes you are intending to model it? Note that if the skew is reasonably small and you do not care very much about it, even if the underlying population is genuinely skewed, then you might still find the normal distribution a useful model to approximate this true distribution of watching times. But you should check that this doesn't end up making silly predictions. Because a normal distribution has no highest or lowest possible value, then although extremely high or low values become increasingly unlikely, you will always find that your model predicts there is some probability of watching for a negative number of hours per day, or more than 24 hours. This gets more problematic for you if the predicted probability of such impossible events becomes high. A symmetric distribution like the normal will predict that as many people will watch for lengths of time more than e.g. 50% above the mean, as watch for less than 50% below the mean. If watching times are very skewed, then this kind of prediction may also be so implausible as to be silly, and give you misleading results if you are taking the results of your model and using them as inputs for some other purpose (for instance, you're running a simulation of watching times in order to calculate optimal advertisement scheduling). If the skewness is so noteworthy you want to capture it as part of your model, then the skew normal distribution may be more appropriate. If you want to capture both skewness and kurtosis, then consider the skewed t. If you want to incorporate the physically possible upper and lower bounds, then consider using the truncated versions of these distributions. Many other probability distributions exist that can be skewed and unimodal (for appropriate parameter choices) such as the F or gamma distributions, and again you can truncate these so they do not predict impossibly high watching times. A beta distribution may be a good choice if you are modelling the fraction of the day spent watching, as this is always bounded between 0 and 1 without further truncation being necessary. If you want to incorporate the concentration of probability at exactly zero due to non-watchers, then consider building in a hurdle model.
But at the point you are trying to throw in every feature you can identify from your data, and build an ever more sophisticated model, perhaps you should ask yourself why you are doing this? Would there be an advantage to a simpler model, for example it being easier to work with mathematically or having fewer parameters to estimate? If you are concerned that such simplification will leave you unable to capture all of the properties of interest to you, it may well be that no "off-the-shelf" distribution does quite what you want. However, we are not restricted to working with named distributions whose mathematical properties have been elucidated previously. Instead, consider using your data to construct an empirical distribution function. This will capture all the behaviour that was present in your data, but you can no longer give it a name like "normal" or "gamma", nor can you apply mathematical properties that pertain only to a particular distribution. For instance, the "95% of the data lies within 1.96 standard deviations of the mean" rule is for normally distributed data and may not apply to your distribution; though note that some rules apply to all distributions, e.g. Chebyshev's inequality guarantees at least 75% of your data must lie within two standard deviations of the mean, regardless of the skew. Unfortunately the empirical distribution will also inherit all those properties of your data set arising purely by sampling error, not just those possessed by the underlying population, so you may find a histogram of your empirical distribution has some humps and dips that the population itself does not. You may want to investigate smoothed empirical distribution functions, or better yet, increasing your sample size.
In summary: although the normal distribution has zero skew, the fact your data are skewed doesn't rule out the normal distribution as a useful model, though it does suggest some other distribution may be more appropriate. You should consider other properties of the data when choosing your model, besides the skew, and consider too the purposes you are going to use the model for. It's safe to say that your true population of watching times does not exactly follow some famous, named distribution, but this does not mean such a distribution is doomed to be useless as a model. However, for some purposes you may prefer to just use the empirical distribution itself, rather than try fitting a standard distribution to it.
$endgroup$
"Is there a better word for that distribution?"
There's a worthwhile distinction here between using words to describe the properties of the distribution, versus trying to find a "name" for the distribution so that you can identify it as (approximately) an instance of a particular standard distribution: one for which a formula or statistical tables might exist for its distribution function, and for which you could estimate its parameters. In this latter case, you are likely using the named distribution, e.g. "normal/Gaussian" (the two terms are generally synonymous), as a model that captures some of the key features of your data, rather than claiming the population your data is drawn from exactly follows that theoretical distribution. To slightly misquote George Box, all models are "wrong", but some are useful. If you are thinking about the modelling approach, it is worth considering what features you want to incorporate and how complicated or parsimonious you want your model to be.
Being positively skewed is an example of describing a property that the distribution has, but doesn't come close to specifying which off-the-shelf distribution is "the" appropriate model. It does rule out some candidates, for example the Gaussian (i.e. normal) distribution has zero skew so will not be appropriate to model your data if the skew is an important feature. There may be other properties of the data that are important to you too, e.g. that it's unimodal (has just one peak) or that it is bounded between 0 and 24 hours (or between 0 and 1, if you are writing it as a fraction of the day), or that there is a probability mass concentrated at zero (since there are people who do not watch youtube at all on a given day). You may also be interested in other properties like the kurtosis. And it is worth bearing in mind that even if your distribution had a "hump" or "bell-curve" shape and had zero or near-zero skew, it doesn't automatically follow that the normal distribution is "correct" for it! On the other hand, even if the population your data is drawn from actually did follow a particular distribution precisely, due to sampling error your dataset may not quite resemble it. Small data sets are likely to be "noisy", and it may be unclear whether certain features you can see, e.g. additional small humps or asymmetric tails, are properties of the underlying population the data was drawn from (and perhaps therefore ought to be incorporated in your model) or whether they are just artefacts from your particular sample (and for modelling purposes should be ignored). If you have a small data set and the skew is close to zero, then it is even plausible the underlying distribution is actually symmetric. The larger your data set and the larger the skewness, the less plausible this becomes — but while you could perform a significance test to see how convincing is the evidence your data provides for skewness in the population it was drawn from, this may be missing the point as to whether a normal (or other zero skew) distribution is appropriate as a model ...
Which properties of the data really matter for the purposes you are intending to model it? Note that if the skew is reasonably small and you do not care very much about it, even if the underlying population is genuinely skewed, then you might still find the normal distribution a useful model to approximate this true distribution of watching times. But you should check that this doesn't end up making silly predictions. Because a normal distribution has no highest or lowest possible value, then although extremely high or low values become increasingly unlikely, you will always find that your model predicts there is some probability of watching for a negative number of hours per day, or more than 24 hours. This gets more problematic for you if the predicted probability of such impossible events becomes high. A symmetric distribution like the normal will predict that as many people will watch for lengths of time more than e.g. 50% above the mean, as watch for less than 50% below the mean. If watching times are very skewed, then this kind of prediction may also be so implausible as to be silly, and give you misleading results if you are taking the results of your model and using them as inputs for some other purpose (for instance, you're running a simulation of watching times in order to calculate optimal advertisement scheduling). If the skewness is so noteworthy you want to capture it as part of your model, then the skew normal distribution may be more appropriate. If you want to capture both skewness and kurtosis, then consider the skewed t. If you want to incorporate the physically possible upper and lower bounds, then consider using the truncated versions of these distributions. Many other probability distributions exist that can be skewed and unimodal (for appropriate parameter choices) such as the F or gamma distributions, and again you can truncate these so they do not predict impossibly high watching times. A beta distribution may be a good choice if you are modelling the fraction of the day spent watching, as this is always bounded between 0 and 1 without further truncation being necessary. If you want to incorporate the concentration of probability at exactly zero due to non-watchers, then consider building in a hurdle model.
But at the point you are trying to throw in every feature you can identify from your data, and build an ever more sophisticated model, perhaps you should ask yourself why you are doing this? Would there be an advantage to a simpler model, for example it being easier to work with mathematically or having fewer parameters to estimate? If you are concerned that such simplification will leave you unable to capture all of the properties of interest to you, it may well be that no "off-the-shelf" distribution does quite what you want. However, we are not restricted to working with named distributions whose mathematical properties have been elucidated previously. Instead, consider using your data to construct an empirical distribution function. This will capture all the behaviour that was present in your data, but you can no longer give it a name like "normal" or "gamma", nor can you apply mathematical properties that pertain only to a particular distribution. For instance, the "95% of the data lies within 1.96 standard deviations of the mean" rule is for normally distributed data and may not apply to your distribution; though note that some rules apply to all distributions, e.g. Chebyshev's inequality guarantees at least 75% of your data must lie within two standard deviations of the mean, regardless of the skew. Unfortunately the empirical distribution will also inherit all those properties of your data set arising purely by sampling error, not just those possessed by the underlying population, so you may find a histogram of your empirical distribution has some humps and dips that the population itself does not. You may want to investigate smoothed empirical distribution functions, or better yet, increasing your sample size.
In summary: although the normal distribution has zero skew, the fact your data are skewed doesn't rule out the normal distribution as a useful model, though it does suggest some other distribution may be more appropriate. You should consider other properties of the data when choosing your model, besides the skew, and consider too the purposes you are going to use the model for. It's safe to say that your true population of watching times does not exactly follow some famous, named distribution, but this does not mean such a distribution is doomed to be useless as a model. However, for some purposes you may prefer to just use the empirical distribution itself, rather than try fitting a standard distribution to it.
edited Apr 2 at 22:15
answered Apr 2 at 1:19
SilverfishSilverfish
15.2k1567147
15.2k1567147
add a comment |
add a comment |
$begingroup$
"Normal" and "Gaussian" mean exactly the same thing. As other answers explain, the distribution you're talking about is not normal/Gaussian, because that distribution assigns probabilities to every value on the real line, whereas your distribution only exists between $0$ and $24$.
$endgroup$
add a comment |
$begingroup$
"Normal" and "Gaussian" mean exactly the same thing. As other answers explain, the distribution you're talking about is not normal/Gaussian, because that distribution assigns probabilities to every value on the real line, whereas your distribution only exists between $0$ and $24$.
$endgroup$
add a comment |
$begingroup$
"Normal" and "Gaussian" mean exactly the same thing. As other answers explain, the distribution you're talking about is not normal/Gaussian, because that distribution assigns probabilities to every value on the real line, whereas your distribution only exists between $0$ and $24$.
$endgroup$
"Normal" and "Gaussian" mean exactly the same thing. As other answers explain, the distribution you're talking about is not normal/Gaussian, because that distribution assigns probabilities to every value on the real line, whereas your distribution only exists between $0$ and $24$.
answered Apr 1 at 14:08
David RicherbyDavid Richerby
1555
1555
add a comment |
add a comment |
$begingroup$
In the case at hand, since the time spent per day is bound from $0$ to $1$ (if quantified as a fraction of the day), distributions that are unbounded above (e.g. Pareto, skew-normal, Gamma, log-normal) won't work, but Beta would.
$endgroup$
add a comment |
$begingroup$
In the case at hand, since the time spent per day is bound from $0$ to $1$ (if quantified as a fraction of the day), distributions that are unbounded above (e.g. Pareto, skew-normal, Gamma, log-normal) won't work, but Beta would.
$endgroup$
add a comment |
$begingroup$
In the case at hand, since the time spent per day is bound from $0$ to $1$ (if quantified as a fraction of the day), distributions that are unbounded above (e.g. Pareto, skew-normal, Gamma, log-normal) won't work, but Beta would.
$endgroup$
In the case at hand, since the time spent per day is bound from $0$ to $1$ (if quantified as a fraction of the day), distributions that are unbounded above (e.g. Pareto, skew-normal, Gamma, log-normal) won't work, but Beta would.
answered Apr 1 at 10:36
J.G.J.G.
27616
27616
add a comment |
add a comment |
$begingroup$
How about a hurdle model?
A hurdle model has two parts. The first is Bernoulli experiment that determines whether you use YouTube at all. If you don't, then your usage time is obviously zero and you're done. If you do, you "pass that hurdle", then the usage time comes from some other strictly positive distribution.
A closely related concept are zero-inflated models. These are meant to deal with a situation where we observe a bunch of zeros, but can't distinguish between always-zeros and sometimes-zeros. For example, consider the number of cigarettes that a person smokes each day. For non-smokers, that number is always zero, but some smokers may not smoke on a given day (out of cigarettes? on a long flight?). Unlike the hurdle model, the "smoker" distribution here should include zero, but these counts are 'inflated' by the non-smokers' contribution too.
$endgroup$
add a comment |
$begingroup$
How about a hurdle model?
A hurdle model has two parts. The first is Bernoulli experiment that determines whether you use YouTube at all. If you don't, then your usage time is obviously zero and you're done. If you do, you "pass that hurdle", then the usage time comes from some other strictly positive distribution.
A closely related concept are zero-inflated models. These are meant to deal with a situation where we observe a bunch of zeros, but can't distinguish between always-zeros and sometimes-zeros. For example, consider the number of cigarettes that a person smokes each day. For non-smokers, that number is always zero, but some smokers may not smoke on a given day (out of cigarettes? on a long flight?). Unlike the hurdle model, the "smoker" distribution here should include zero, but these counts are 'inflated' by the non-smokers' contribution too.
$endgroup$
add a comment |
$begingroup$
How about a hurdle model?
A hurdle model has two parts. The first is Bernoulli experiment that determines whether you use YouTube at all. If you don't, then your usage time is obviously zero and you're done. If you do, you "pass that hurdle", then the usage time comes from some other strictly positive distribution.
A closely related concept are zero-inflated models. These are meant to deal with a situation where we observe a bunch of zeros, but can't distinguish between always-zeros and sometimes-zeros. For example, consider the number of cigarettes that a person smokes each day. For non-smokers, that number is always zero, but some smokers may not smoke on a given day (out of cigarettes? on a long flight?). Unlike the hurdle model, the "smoker" distribution here should include zero, but these counts are 'inflated' by the non-smokers' contribution too.
$endgroup$
How about a hurdle model?
A hurdle model has two parts. The first is Bernoulli experiment that determines whether you use YouTube at all. If you don't, then your usage time is obviously zero and you're done. If you do, you "pass that hurdle", then the usage time comes from some other strictly positive distribution.
A closely related concept are zero-inflated models. These are meant to deal with a situation where we observe a bunch of zeros, but can't distinguish between always-zeros and sometimes-zeros. For example, consider the number of cigarettes that a person smokes each day. For non-smokers, that number is always zero, but some smokers may not smoke on a given day (out of cigarettes? on a long flight?). Unlike the hurdle model, the "smoker" distribution here should include zero, but these counts are 'inflated' by the non-smokers' contribution too.
answered Apr 1 at 13:58
Matt KrauseMatt Krause
15k24480
15k24480
add a comment |
add a comment |
$begingroup$
If the distribution is indeed a 'subset' of the normal distribution, you should considder a truncated model. Widely used in this context is the family of TOBIT models.
They essentialy suggest a pdf with a (positive) probability mass at 0 and then a 'cut of part of the normal distribution' for positive values.
I will refrain from typing the formula here and rather refere you to the Wikipedia Article: https://en.wikipedia.org/wiki/Tobit_model
$endgroup$
add a comment |
$begingroup$
If the distribution is indeed a 'subset' of the normal distribution, you should considder a truncated model. Widely used in this context is the family of TOBIT models.
They essentialy suggest a pdf with a (positive) probability mass at 0 and then a 'cut of part of the normal distribution' for positive values.
I will refrain from typing the formula here and rather refere you to the Wikipedia Article: https://en.wikipedia.org/wiki/Tobit_model
$endgroup$
add a comment |
$begingroup$
If the distribution is indeed a 'subset' of the normal distribution, you should considder a truncated model. Widely used in this context is the family of TOBIT models.
They essentialy suggest a pdf with a (positive) probability mass at 0 and then a 'cut of part of the normal distribution' for positive values.
I will refrain from typing the formula here and rather refere you to the Wikipedia Article: https://en.wikipedia.org/wiki/Tobit_model
$endgroup$
If the distribution is indeed a 'subset' of the normal distribution, you should considder a truncated model. Widely used in this context is the family of TOBIT models.
They essentialy suggest a pdf with a (positive) probability mass at 0 and then a 'cut of part of the normal distribution' for positive values.
I will refrain from typing the formula here and rather refere you to the Wikipedia Article: https://en.wikipedia.org/wiki/Tobit_model
answered Apr 2 at 12:25
LucasLucas
101
101
add a comment |
add a comment |
$begingroup$
Normal distributions are by definition non-skewed, so you can't have both things. If the distribution is left-skewed, then it cannot be Gaussian. You'll have to pick a different one! The closest thing to your request I can think of is this:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
add a comment |
$begingroup$
Normal distributions are by definition non-skewed, so you can't have both things. If the distribution is left-skewed, then it cannot be Gaussian. You'll have to pick a different one! The closest thing to your request I can think of is this:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
add a comment |
$begingroup$
Normal distributions are by definition non-skewed, so you can't have both things. If the distribution is left-skewed, then it cannot be Gaussian. You'll have to pick a different one! The closest thing to your request I can think of is this:
https://en.wikipedia.org/wiki/Skew_normal_distribution
$endgroup$
Normal distributions are by definition non-skewed, so you can't have both things. If the distribution is left-skewed, then it cannot be Gaussian. You'll have to pick a different one! The closest thing to your request I can think of is this:
https://en.wikipedia.org/wiki/Skew_normal_distribution
answered Apr 2 at 7:59
DavidDavid
4955
4955
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
add a comment |
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
3
3
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
I agree except that the OP is confusing left and right skewness, as already pointed out. And @behold has already suggested the skew-normal in an answer. So, I can't see that this adds to existing answers.
$endgroup$
– Nick Cox
Apr 2 at 9:48
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
$begingroup$
It summarizes many of them in a straight-forward three-line response
$endgroup$
– David
Apr 2 at 11:46
3
3
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
Sorry, but that's still repetition.
$endgroup$
– Nick Cox
Apr 2 at 12:52
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
$begingroup$
OK... who cares?
$endgroup$
– David
Apr 2 at 14:06
2
2
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
$begingroup$
Well, I do; and whoever added +1 to my comments (clearly not me) and whoever downvoted your answer (not me, as it happens). This thread is already long and repetitive; yet more redundant comments don't improve it for future readers.
$endgroup$
– Nick Cox
Apr 2 at 14:24
add a comment |
protected by gung♦ Apr 2 at 13:16
Thank you for your interest in this question.
Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).
Would you like to answer one of these unanswered questions instead?
4
$begingroup$
As some answers mention but do not emphasise, skewness is named informally for the longer tail if there is one, so right-skewed if a longer right tail. Left and right as used in this context both presuppose a display following a convention that magnitude is shown on the hoirizontal axis. If that sounds too obvious, consider displays in the Earth and environmental sciences in which the magnitude is height or depth and shown vertically. Small print: some measures of skewness can be zero even if a distribution is skewed geometrically.
$endgroup$
– Nick Cox
Mar 31 at 6:42
1
$begingroup$
Total time per day for all users? or time per day per person? If the latter, then surely there's a moderately big spike at 0, in which case you probably need a 'spike and slab' style distribution with a Dirac delta at 0.
$endgroup$
– innisfree
Apr 1 at 7:08
5
$begingroup$
"Normal" is synonymous with "Gaussian", and Gaussian distributions, also called normal distributions, are not skewed.
$endgroup$
– Michael Hardy
Apr 2 at 1:47
$begingroup$
I find the question in the title much different from the question in the body text. Or at least the title is very confusing. No distribution is 'normal but highly skewed' that's a contradiction. Also, the Gaussian distribution is very well defined $f(x) = frac1sqrt2pisigma^2 textexpleft( - frac(x-mu)^22sigma^2right)$ and not at all like the distribution of time spent per day on YouTube. So the answer to the question in the title is a big no.
$endgroup$
– Martijn Weterings
Apr 2 at 12:39
2
$begingroup$
also, the question at the end 'is there a better word for that distribution?' is very vague or broad. The information seems to be only 'one mode' and 'a long right tail' (the part 'probably normally distributed' makes no sense). There can be many distributions that satisfy these conditions. It is amazing that this question attracts more than ten answers and at least as many proposals for the alternative distribution before we actually try to clarify the question (there isn't even data).
$endgroup$
– Martijn Weterings
Apr 2 at 12:53