[FIXED] How to match labels and their cluster point in k-means clusters?

Issue

I am doing a k-means clustering and I want to make sure that the labels are matched to the correct cluster number. Below is the code I used

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from sklearn.metrics import pairwise_distances_argmin_min

dataset = pd.read_csv('ratio.csv', index_col=0).T
dataset_copy = dataset
dataset_copy = dataset_copy.dropna()

X = dataset_copy.iloc[:, [0, 1, 2, 3]].values

kmeans = KMeans(n_clusters=4, init='k-means++', random_state=42)
y_kmeans = kmeans.fit_predict(X)

# From here
Company = pd.DataFrame(dataset_copy.index)
cluster_labels = pd.DataFrame(kmeans.labels_)

labels_df = pd.concat([Company, cluster_labels],axis = 1)

Does the code from the # From here assign the correct label to the cluster number?

Part of the dataset that I am using in the code:

Solution

Yes, the code snippet you provided from the "# From here" comment does correctly assign the cluster labels to each company.

The relevant lines of code:

Company = pd.DataFrame(dataset_copy.index): This line creates a DataFrame from the indices of dataset_copy, which presumably are the company names or identifiers.

cluster_labels = pd.DataFrame(kmeans.labels_): Here, you are converting the labels assigned by the k-means algorithm to a DataFrame. The kmeans.labels_ array contains the cluster number assigned to each sample in X.

labels_df = pd.concat([Company, cluster_labels], axis=1): This line concatenates the company names and their corresponding cluster labels along the columns (axis=1). This results in a new DataFrame labels_df, where each row contains a company name and its associated cluster label.

Thus, every company in labels_df is matched with the cluster number assigned by the k-means algorithm.

However, there is a more elegant way. Both dataset_copy.index and kmeans.labels_ are 1D arrays, so you could create the DataFrame for mapping in a more elegant way:

labels_df = pd.DataFrame({'Company': dataset_copy.index, 'Cluster': kmeans.labels_})

Answered By - DataJanitor