Trjtdtk

How would a mousetrap for use in space work?

How to write this math term? with cases it isn't working

When a candle burns, why does the top of wick glow if bottom of flame is hottest?

Performance gap between bool std:vector and array

What do you call the main part of a joke?

Fundamental Solution of the Pell Equation

Amount of permutations on an NxNxN Rubik's Cube

Maximum summed subsequences with non-adjacent items

Can anything be seen from the center of the Boötes void? How dark would it be?

Disembodied hand growing fangs

Selecting user stories during sprint planning

Withdrew £2800, but only £2000 shows as withdrawn on online banking; what are my obligations?

Redirect to div in page with #

How do I make this wiring inside cabinet safer?

Project Euler #1 in C++

Is CEO the "profession" with the most psychopaths?

How do I use the new nonlinear finite element in Mathematica 12 for this equation?

Is there hard evidence that the grant peer review system performs significantly better than random?

If a VARCHAR(MAX) column is included in an index, is the entire value always stored in the index page(s)?

What's the meaning of "fortified infraction restraint"?

Is it ethical to give a final exam after the professor has quit before teaching the remaining chapters of the course?

How could we fake a moon landing now?

Is there a kind of relay only consumes power when switching?

How to Make a Beautiful Stacked 3D Plot

Discrete Ordinal Classification with Probabilities

0

$begingroup$

If I have classes 1, 2, 3 and 4. But, I also need the probability for each of the other classes. I'm currently using XGBoost for one-vs-rest classification, but that means we're losing information since the classes are ordinal.

machine-learning xgboost

share|improve this question

edited Apr 3 at 6:02

winnie

asked Apr 2 at 19:38

winniewinnie

11

$endgroup$

• 
  
  
  

  
  1
  

  
  
  
  
  
  

  
  $begingroup$
  can you add a bit more detail to what your question is? if you're just looking for a model for ordinal regression a quick internet search can get you the answer
  $endgroup$
  – oW_♦
  Apr 2 at 20:04
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @oW_, most ordinal regression only gives the final prediction. I'd prefer a probably for each of 25 discrete classes. So, I need a probably that the y_pred == 1, y_pred == 2 etc.
  $endgroup$
  – winnie
  Apr 3 at 6:00
  

  
  
  
  

add a comment | 

0

$begingroup$

If I have classes 1, 2, 3 and 4. But, I also need the probability for each of the other classes. I'm currently using XGBoost for one-vs-rest classification, but that means we're losing information since the classes are ordinal.

machine-learning xgboost

share|improve this question

edited Apr 3 at 6:02

winnie

asked Apr 2 at 19:38

winniewinnie

11

$endgroup$

• 
  
  
  

  
  1
  

  
  
  
  
  
  

  
  $begingroup$
  can you add a bit more detail to what your question is? if you're just looking for a model for ordinal regression a quick internet search can get you the answer
  $endgroup$
  – oW_♦
  Apr 2 at 20:04
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @oW_, most ordinal regression only gives the final prediction. I'd prefer a probably for each of 25 discrete classes. So, I need a probably that the y_pred == 1, y_pred == 2 etc.
  $endgroup$
  – winnie
  Apr 3 at 6:00
  

  
  
  
  

add a comment | 

$begingroup$

If I have classes 1, 2, 3 and 4. But, I also need the probability for each of the other classes. I'm currently using XGBoost for one-vs-rest classification, but that means we're losing information since the classes are ordinal.

machine-learning xgboost

share|improve this question

edited Apr 3 at 6:02

winnie

asked Apr 2 at 19:38

winniewinnie

11

$endgroup$

share|improve this question

edited Apr 3 at 6:02

winnie

asked Apr 2 at 19:38

winniewinnie

11

share|improve this question

edited Apr 3 at 6:02

winnie

asked Apr 2 at 19:38

winniewinnie

11

asked Apr 2 at 19:38

winniewinnie

11

asked Apr 2 at 19:38

winniewinnie

11

2 Answers
2

active

oldest

votes

0

$begingroup$

I think for obtaining probabilities for each of the classes you should be doing multiclass classification, instead of one-vs-rest classification. In one-vs-rest classification, the rest of the classes grouped together as one class so getting probabilities for each separately is not possible.

share|improve this answer

answered Apr 3 at 6:14

tamtam

1014

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  I can get the probabilities from one-vs-rest just by getting the one-vs-rest probability for each of the classes. But, it's losing the fact that output is ordinal, is losing vital information.
  $endgroup$
  – winnie
  Apr 3 at 6:19
  

  
  
  
  

add a comment | 

0

$begingroup$

Probit and logit ordinal regression model the cumulative probabilities $P(y leq i | x)$. So you should be able to get class probabilities by taking the difference $P(y=i |x) = P(y leq i|x) - P(yleq i-1 |x)$. Then it just depends on the implementation whether or not it makes the probabilities available to you.

In R, for example, you can get these probabilities from the polr package.

[Even though you lose some information when doing one-vs-rest classification, ordinal regression models make some proportionality assumptions that (if not appropriate for your data) may not necessarily give you better results.]

share|improve this answer

answered Apr 3 at 15:26

oW_♦oW_

3,381933

$endgroup$

add a comment | 

Your Answer

StackExchange.ready(function()
var channelOptions =
tags: "".split(" "),
id: "557"
;
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function()
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled)
StackExchange.using("snippets", function()
createEditor();
);

else
createEditor();

);

function createEditor()
StackExchange.prepareEditor(
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
bindNavPrevention: true,
postfix: "",
imageUploader:
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
,
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
);



);

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);

Sign up using Google

Sign up using Facebook

Sign up using Email and Password

Post as a guest

Name

Email

Required, but never shown

StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f48467%2fdiscrete-ordinal-classification-with-probabilities%23new-answer', 'question_page');

);

Post as a guest

Name

Email

Required, but never shown

0

$begingroup$

I think for obtaining probabilities for each of the classes you should be doing multiclass classification, instead of one-vs-rest classification. In one-vs-rest classification, the rest of the classes grouped together as one class so getting probabilities for each separately is not possible.

share|improve this answer

answered Apr 3 at 6:14

tamtam

1014

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  I can get the probabilities from one-vs-rest just by getting the one-vs-rest probability for each of the classes. But, it's losing the fact that output is ordinal, is losing vital information.
  $endgroup$
  – winnie
  Apr 3 at 6:19
  

  
  
  
  

add a comment | 

0

$begingroup$

I think for obtaining probabilities for each of the classes you should be doing multiclass classification, instead of one-vs-rest classification. In one-vs-rest classification, the rest of the classes grouped together as one class so getting probabilities for each separately is not possible.

share|improve this answer

answered Apr 3 at 6:14

tamtam

1014

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  I can get the probabilities from one-vs-rest just by getting the one-vs-rest probability for each of the classes. But, it's losing the fact that output is ordinal, is losing vital information.
  $endgroup$
  – winnie
  Apr 3 at 6:19
  

  
  
  
  

add a comment | 

$begingroup$

I think for obtaining probabilities for each of the classes you should be doing multiclass classification, instead of one-vs-rest classification. In one-vs-rest classification, the rest of the classes grouped together as one class so getting probabilities for each separately is not possible.

share|improve this answer

answered Apr 3 at 6:14

tamtam

1014

$endgroup$

share|improve this answer

answered Apr 3 at 6:14

tamtam

1014

answered Apr 3 at 6:14

tamtam

1014

answered Apr 3 at 6:14

tamtam

1014

0

$begingroup$

Probit and logit ordinal regression model the cumulative probabilities $P(y leq i | x)$. So you should be able to get class probabilities by taking the difference $P(y=i |x) = P(y leq i|x) - P(yleq i-1 |x)$. Then it just depends on the implementation whether or not it makes the probabilities available to you.

In R, for example, you can get these probabilities from the polr package.

[Even though you lose some information when doing one-vs-rest classification, ordinal regression models make some proportionality assumptions that (if not appropriate for your data) may not necessarily give you better results.]

share|improve this answer

answered Apr 3 at 15:26

oW_♦oW_

3,381933

$endgroup$

add a comment | 

0

$begingroup$

Probit and logit ordinal regression model the cumulative probabilities $P(y leq i | x)$. So you should be able to get class probabilities by taking the difference $P(y=i |x) = P(y leq i|x) - P(yleq i-1 |x)$. Then it just depends on the implementation whether or not it makes the probabilities available to you.

In R, for example, you can get these probabilities from the polr package.

[Even though you lose some information when doing one-vs-rest classification, ordinal regression models make some proportionality assumptions that (if not appropriate for your data) may not necessarily give you better results.]

share|improve this answer

answered Apr 3 at 15:26

oW_♦oW_

3,381933

$endgroup$

add a comment | 

$begingroup$

Probit and logit ordinal regression model the cumulative probabilities $P(y leq i | x)$. So you should be able to get class probabilities by taking the difference $P(y=i |x) = P(y leq i|x) - P(yleq i-1 |x)$. Then it just depends on the implementation whether or not it makes the probabilities available to you.

In R, for example, you can get these probabilities from the polr package.

[Even though you lose some information when doing one-vs-rest classification, ordinal regression models make some proportionality assumptions that (if not appropriate for your data) may not necessarily give you better results.]

share|improve this answer

answered Apr 3 at 15:26

oW_♦oW_

3,381933

$endgroup$

share|improve this answer

answered Apr 3 at 15:26

oW_♦oW_

3,381933

answered Apr 3 at 15:26

oW_♦oW_

3,381933

answered Apr 3 at 15:26

oW_♦oW_

3,381933