cross_val_predict is not producing the same number of variables after transforming data with a Pipeline

Question

I\'m running into an value error when using cross_val_predict(). For my project I want to test the performance of my model based on different values of k in Sel