def top_features(df, features, top_n, y_var):
"""
Returns top_n features based on F-values from ANOVA.
"""
# Get X and y from df
X , y = X_y_splitter(df, features, y_var)
f_values = SelectKBest(f_classif, k=top_n).fit(X, y).scores_
f_values = pd.DataFrame(
{
'feature': X.columns,
'f_values': f_values
}
)
f_values.sort_values('f_values', ascending=False, inplace=True)
f_values.head()
top_n_features = f_values['feature'][0:top_n].tolist()
return(top_n_features)
