compare.py

def compare():
    for is_le in [True, False]:
        method = 'label encoder'
        
        if is_le:
            selected = df_le[selects_le + ['is_canceled']]
        else:
            selected = df_hot[selects_hot + ['is_canceled']]
            method = 'dummy variables'
        
        # separate majority and minority classes
        major = selected[selected['is_canceled'] == 0]
        minor = selected[selected['is_canceled'] == 1]

        # downsample majority class
        downsampled = resample(major, replace=False, n_samples=len(minor), random_state=123) 

        # combine minority class with downsampled majority class
        df_new = pd.concat([downsampled, minor])
    
        X = df_new.drop('is_canceled', axis=1)
        y = df_new['is_canceled']

        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.2, random_state=42)

        log = LogisticRegression().fit(X_train, y_train)
        y_pred = log.predict(X_test)

        print(f'Accuracy for {method}: {accuracy_score(y_test, y_pred)}')
        print(f'Classification report for {method}:\n{classification_report(y_test, y_pred)}')