BookmarkSubscribeRSS Feed
AI8
Calcite | Level 5 AI8
Calcite | Level 5

Hi! I am constructing an ordinal stepwise logistic regression model in SAS Enterprise Miner. I am working with a Human Resource Management data set containing employee survey data. My target variable is "Performance Rating" which is an ordinal target with 2 levels "3" meaning that performance is good and "4" meaning performance is outstanding. I am having issues with making an ordinal regression model with this variable and when I use selection criterion as validation misclassification rate or Validation error it doesn't produce an output. 

 

I would appreciate any helpful feedback.

https://www.kaggle.com/vjchoudhary7/hr-analytics-case-study

 

Logistic Regression OutputLogistic Regression OutputVariable Levels and RolesVariable Levels and RolesLogistic Regression PropertiesLogistic Regression Properties

1 REPLY 1
sbxkoenk
SAS Super FREQ

Hello,

 

I think Enterprise Miner does produce output but your final model is an intercept-only model (one effect!). None of the candidate predictors was retained (when using Validation Misclassification as selection criterion).

You should try to make a better model (maybe by transforming candidate predictors).

Also, if your ordinal target just has two levels, you can just as well mark your target as having the binary measurement scale.

 

It has been a long time since I have used Enterprise Miner (I use Model Studio VDMML nowadays) but I think EM Logistic Regression node cannot even deal with ordinal targets.

For models with a multinomial response (> 2 levels) you need "Generalized Logits". I think that's available.

But for models with an ordinal response (> 2 levels) you need cumulative logit, cumulative probit, or cumulative complementary log-log links. I don't think these are available.

 

Good luck,

Koen

sas-innovate-2024.png

Available on demand!

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.

 

Register now!

How to choose a machine learning algorithm

Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 1028 views
  • 0 likes
  • 2 in conversation