drop_col_feat_imp.py

from sklearn.base import clone 

def drop_col_feat_imp(model, X_train, y_train, random_state = 42):
    
    # clone the model to have the exact same specification as the one initially trained
    model_clone = clone(model)
    # set random_state for comparability
    model_clone.random_state = random_state
    # training and scoring the benchmark model
    model_clone.fit(X_train, y_train)
    benchmark_score = model_clone.score(X_train, y_train)
    # list for storing feature importances
    importances = []
    
    # iterating over all columns and storing feature importance (difference between benchmark and new model)
    for col in X_train.columns:
        model_clone = clone(model)
        model_clone.random_state = random_state
        model_clone.fit(X_train.drop(col, axis = 1), y_train)
        drop_col_score = model_clone.score(X_train.drop(col, axis = 1), y_train)
        importances.append(benchmark_score - drop_col_score)
    
    importances_df = imp_df(X_train.columns, importances)
    return importances_df

imp_df(X_train.columns, importances)?
imp_df is not defined!

cannot find imp_df

This gist was not supposed to be a standalone function, but was created to show the logic in the article. imp_df can be found in the notebook hosting entire code used for the article: https://github.com/erykml/medium_articles/blob/master/feature_importance.ipynb

function for creating a feature importance dataframe

def imp_df(column_names, importances):
df = pd.DataFrame({'feature': column_names,
'feature_importance': importances})
.sort_values('feature_importance', ascending = False)
.reset_index(drop = True)
return df

I was wondering why you are fitting and scoring on X_train, y_train. Will this not give a score of 1 for each column?