RFM Streamlit

rfm_streamlit.py

def rfm_segmentation(df, technique:int, num_segments:int):
    if technique == 1:
        data = df.copy()
        # K-means clustering
        X_ = data[['recency', 'frequency', 'monetary']]
        X = StandardScaler().fit_transform(X_)
        kms = KMeansInterp(
	    n_clusters=num_segments, random_state=42,
	    ordered_feature_names=X_.columns.tolist(), 
	    n_init = 'auto', max_iter = 1000,
	    feature_importance_method='wcss_min', # or 'unsup2sup'
        ).fit(X)
        #kmeans = KMeans(n_clusters=num_segments, random_state=42, n_init='auto', max_iter=1000).fit(X)
        segment = kms.labels_ + 1
        data['RFM_Segment'] = segment
        data['RFM_Segment'] = data['RFM_Segment'].astype(str)
        feature_importance = kms.feature_importances_
        return data, feature_importance