Because disposition variables are directly related to the outcome, we cannot leave them in the data. We omit them here (recall that we previously removed several of the disposition variables right after we created our final target column):
disp_cols = [
'NODISP','NOFU','RETRNED','RETREFFU','LEFTBTRI',
'LEFTAMA','DOA','DIEDED','TRANNH','OTHDISP',
'ADMIT','ADMTPHYS','BOARDED','LOS','HDDIAG1',
'HDDIAG2','HDDIAG3','HDDIAG1R','HDDIAG2R','HDDIAG3R',
'HDSTAT','ADISP','OBSSTAY','STAY24'
]
X_train.drop(disp_cols, axis=1, inplace=True)
X_test.drop(disp_cols, axis=1, inplace=True)