This gist was made to demonstrate code on medium.

rf-explanation4.py

def auto_applyer(leaf_value,feature_value,feature_list,df_raw1,df_test,target_column,date_column=None):
    reset_rf_samples()

    if date_column:
        if date_column in df_test:
            add_datepart(df_test,date_column)
        if date_column in df_raw1:
            add_datepart(df_raw1,date_column)
    '''First we will pre process both test and raw data'''
    train_cats(df_raw1)
    apply_cats(df=df_test,trn=df_raw1)
    X,y,nas=proc_df(df_raw1,target_column)
    X_test,_,nas = proc_df(df_test, na_dict=nas)
    X=X[feature_list]
    X_test=X_test[feature_list]
    z=RandomForestRegressor(n_jobs=-1,min_samples_leaf=leaf_value,max_features=feature_value,oob_score=False,n_estimators=75)
    z.fit(X,y)
    fi = rf_feat_importance(z,X)
    graphed=fi.plot('cols', 'imp', 'barh', figsize=(12,7), legend=False)
    fig_save = graphed.get_figure()
    fig_save.savefig('Feature Importance.png')
    print(z.predict(X_test))
    return z.predict(X_test)