Model Probability Calibration in Pyspark

Question

I am using PySpark to implement a Churn classification model for a business problem and the dataset I have is imbalanced. So when I train the model, I randomly select a dataset