two_sigma_financial_modelling.py 文件源码

def feature_agglomeration(df, number_of_clusters=int(df.shape[1] / 1.2)):
        df = df.copy()
        # Todo: find optimal number of clusters for the feature clustering
        # number_of_clusters = int(df.shape[1]/2)

        agglomerated_features = FeatureAgglomeration(n_clusters=number_of_clusters)
        if any(tuple(df.columns == 'Call Outcome')):
            res = agglomerated_features.fit_transform(np.reshape(np.array(df.dropna().values), df.dropna()
                                                                 .shape), y=df['Call Outcome'].values)
        else:
            res = agglomerated_features.fit_transform(np.reshape(np.array(df.values), df.shape))
        df = pd.DataFrame(data=res)
        return df