A key challenge for e-commerce businesses is to analyze the trend in the market to increase their sales. The trend can be easily observed if the companies can group the customers; based on their activity on the e-commerce site. This grouping can be done by applying different criteria like previous orders, mostly searched brands and so on.

Given the e-commerce data, use k-means clustering algorithm to cluster customers with similar interest.

Remaining 35 features (brands) contains the number of times customers have searched them.

• First Import the necessary libraries needed for the project. Below are the libraries that I have used
  • import pandas as pd
  • import numpy as np
  • import matplotlib.pyplot as plt
  • import seaborn as sns
  • from sklearn.preprocessing import MinMaxScaler
  • from sklearn.metrics import silhouette_score
  • from sklearn.cluster import KMeans
  • from yellowbrick.cluster import KElbowVisualizer
• Load the dataset using pandas.
• Understand the data with basic statistics, info, total records, features and their data types also number of null values.
• Perform the data cleaning techniques like treating the missing values.
• Visualize the data with the help of Matplotlib or Seaborn to get better understanding of the features.
  • In my project I have used Seaborn’s
  • Boxplot(To visualize the Outliers present in the data)
  • Barplot (To get to 10 customer ID based on Total searches)
  • Countplot (To get gender count and Prior orders count)
  • Heatmap (To find the Correlation)
  • Histogram (To get know how the data is distributed)
  • Below are some of the charts that I have plotted

• Perform the data scaling using minmaxscaler and data will be scaled between 0 and 1.
• Implement the Elbow method or Silhouette method to find the optimum number of K. Also visualize them with plots.

• After getting optimum K value use it as number of clusters and perform the model fit.
• The model will provide the output with different clusters.
• Analyed each clusters using visualization techniques to find the insights.

• Finally I got the conclusion (Refer attached python notebook) by analyzing each clusters.