kmeans.py

import pandas as pd
from sklearn.cluster import KMeans
import plotly.express as px

def k_means(dataset, cols, drop_features, sample_data):
    X = sample_data
    print(X)
    N = len(sample_data.columns)
    print(N)

    distortions = []
    K = range(1,11)
    print('ok')

    for i in K:
        try:
            print(i)
            kmeans = KMeans(n_clusters=i, init='k-means++')
            print("length before",len(X.columns))
            kmeans = kmeans.fit(X)
            print("length after fit",len(X.columns))
            distortions.append(kmeans.inertia_)
        except Exception as e:
            print(e)
            pass

    print(distortions)
    df = pd.DataFrame({'Clusters': K, 'Distortions': distortions})
    print(df)

    elbow_curve = (px.line(df, x='Clusters', y='Distortions')).update_traces(mode='lines+markers')

    

    #Silhouette score
#     silhouette_scores = []
#     rang = range(2,12)

#     for cluster_size in rang:
#         kmeans = cluster.KMeans(n_clusters=cluster_size, init='k-means++', random_state=200)
#         labels = kmeans.fit(X).labels_
#         silhouette_score = metrics.silhouette_score(sample_data, 
#                                                     labels, 
#                                                     metric='euclidean', 
#                                                     sample_size=1000, 
#                                                     random_state=200)

#         silhouette_scores.append(silhouette_score)

#     df = pd.DataFrame({'Clusters': rang, 'Silhouette Score': silhouette_scores})
#     silhouette = (px.line(df, x='Clusters', y='Silhouette Score', template='seaborn')).update_traces(mode='lines+markers')
    
    return elbow_curve