0

I am building a logistic regression with Churn as the dependent variable. I cleaned my data and transformed the categorical data to factors. This is the description of the data after updating the categorical data: Data Summary click HERE I then attempted to fit the model to the data but received the error messages "algorithm did not converge" and "fitted probabilities 0 or 1 occurred." See below image: model summary click HERE

I am a newbie to R so am open to feedback. Is there a command/function I can run (in R) to fix this? Are there categorical variables I should not have classified as factors? Please help! Thank you!

  • Your model is probably too complex to be supported by your data. Maybe try lasso or elastic net? Have you checked correlations between your predictors? Have you checked how balanced your dataset is? – Roland Oct 06 '22 at 06:04
  • Hey @Roland, the data is imbalanced, how do you fix that? Is there a function you can run to fix? –  Oct 06 '22 at 06:32
  • You cannot fix imbalanced data. You can only deal with it (if the imbalance causes issues). – Roland Oct 06 '22 at 06:34
  • Thanks @Roland. Doing some research, and I think there are sampling techniques that can address imbalanced data. I am looking into SMOTE now –  Oct 06 '22 at 07:25
  • See https://stats.stackexchange.com/search?q=fitted+probabilities+0+or+1+occurred – whuber Oct 07 '22 at 01:29

0 Answers0