Churn Prediction Model

AI Discussions
,
1

I am creating a churn prediction model, that has a data split of 80% negative, and 20% positive. I have trained a neural network, as well as several boosted decision tree models, however I continue to have the issue where the models are predicting about 95% negative predictions. I have tried adding batch normalization, dropout, and different sampling techniques. Does anyone else have any other suggestions?

2

Have you seen this ?

3

If you can share the data and data model(s) ( May be a GitHub link etc), I can review and may help you with observation(s).

4

Here is the link for the github repo: GitHub - eannmckasson1/ChurnPrediction

1 Like