Trjtdtk

Unexpected email from Yorkshire Bank

How to back up a running remote server?

Reverse the word in a string with the same order in javascript

Minimum value of 4 digit number divided by sum of its digits

Is it possible to Ready a spell to be cast just before the start of your next turn by having the trigger be an ally's attack?

Feels like I am getting dragged in office politics

In gnome-terminal only 2 out of 3 zoom keys work

Did Henry V’s archers at Agincourt fight with no pants / breeches on because of dysentery?

Examples of non trivial equivalence relations , I mean equivalence relations without the expression " same ... as" in their definition?

Is it possible to measure lightning discharges as Nikola Tesla?

Why does Bran Stark feel that Jon Snow "needs to know" about his lineage?

Was it really necessary for the Lunar Module to have 2 stages?

Illegal assignment from SObject to Contact

Where does the labelling of extrinsic semiconductors as "n" and "p" come from?

Why was Germany not as successful as other Europeans in establishing overseas colonies?

What does YCWCYODFTRFDTY mean?

Toggle Overlays shortcut?

Has any spacecraft ever had the ability to directly communicate with civilian air traffic control?

What's the polite way to say "I need to urinate"?

You look catfish vs You look like a catfish

Lock in SQL Server and Oracle

Can fracking help reduce CO2?

Help, my Death Star suffers from Kessler syndrome!

Will tsunami waves travel forever if there was no land?

Finding optimal region of a continuous search space

Tuning Gradient Boosted Classifier's hyperparametrs and balancing itOptimal parameter estimation for a classifier with multiple parametersSales Dataset to determine best model for predicting future salesSpace complexity of classification algorithmsHow to (better) discretize continuous data in decision trees?Non-convex feasible region in a linear programming problemHow does binary cross entropy work?Why do we need a gain ratioSciKit-Learn Decision Tree Overfitting

0

$begingroup$

I'm working on a problem in which I've got large table of data with two features and three target columns and I need to find the optimal filter parameters which theoretically would give me:

1) most preferable target distribution, i.e. highest mean & smallest std on positive target columns and the lowest mean and smallest std on negative target columns

2) largest subset possible

the filtering is done on 2 columns only, so four possible parameters, i.e. min/max of each column and there are three columns which determine the 'character' of the target distribution

since the data is already tabular and continuous, I've just set up a grid search and then I scan for the global max, but as I'm doing this in 2D I can only evaluate on one of the three columns which determine actual character of the target distribution. evaluation of the results that this generates are less than optimal.

I have a sense that there is a much better way to formulate the problem, e.g. with decision trees or a probabilistic model perhaps, but I'm at a loss as to how to go about it. and as I'm new to this, I'm not sure even how to research the issue.

Anybody have direction or suggestions about how to better approach this?

decision-trees optimization probabilistic-programming

share|improve this question

asked Apr 9 at 8:57

JedJed

63

$endgroup$

add a comment | 

0

$begingroup$

I'm working on a problem in which I've got large table of data with two features and three target columns and I need to find the optimal filter parameters which theoretically would give me:

1) most preferable target distribution, i.e. highest mean & smallest std on positive target columns and the lowest mean and smallest std on negative target columns

2) largest subset possible

the filtering is done on 2 columns only, so four possible parameters, i.e. min/max of each column and there are three columns which determine the 'character' of the target distribution

since the data is already tabular and continuous, I've just set up a grid search and then I scan for the global max, but as I'm doing this in 2D I can only evaluate on one of the three columns which determine actual character of the target distribution. evaluation of the results that this generates are less than optimal.

I have a sense that there is a much better way to formulate the problem, e.g. with decision trees or a probabilistic model perhaps, but I'm at a loss as to how to go about it. and as I'm new to this, I'm not sure even how to research the issue.

Anybody have direction or suggestions about how to better approach this?

decision-trees optimization probabilistic-programming

share|improve this question

asked Apr 9 at 8:57

JedJed

63

$endgroup$

add a comment | 

$begingroup$

I'm working on a problem in which I've got large table of data with two features and three target columns and I need to find the optimal filter parameters which theoretically would give me:

1) most preferable target distribution, i.e. highest mean & smallest std on positive target columns and the lowest mean and smallest std on negative target columns

2) largest subset possible

the filtering is done on 2 columns only, so four possible parameters, i.e. min/max of each column and there are three columns which determine the 'character' of the target distribution

since the data is already tabular and continuous, I've just set up a grid search and then I scan for the global max, but as I'm doing this in 2D I can only evaluate on one of the three columns which determine actual character of the target distribution. evaluation of the results that this generates are less than optimal.

I have a sense that there is a much better way to formulate the problem, e.g. with decision trees or a probabilistic model perhaps, but I'm at a loss as to how to go about it. and as I'm new to this, I'm not sure even how to research the issue.

Anybody have direction or suggestions about how to better approach this?

decision-trees optimization probabilistic-programming

share|improve this question

asked Apr 9 at 8:57

JedJed

63

$endgroup$

share|improve this question

asked Apr 9 at 8:57

JedJed

63

share|improve this question

asked Apr 9 at 8:57

JedJed

63

asked Apr 9 at 8:57

JedJed

63

asked Apr 9 at 8:57

JedJed

63