AIB503 ECA (End-of-Course Assessment) SUSS: July Semester 2024 – Foundation to Python for AI, Singapore

Question 1

Clustering Analysis on Customer Profiles

This assignment is to utilize Python for applying clustering techniques to group customers based on their profiles, enabling the tailoring of marketing strategies and services. It provides an efficient method for organizing and managing the company’s customer relationships, ultimately boosting customer loyalty and conversions.

The dataset contains the following variables. https://www.kaggle.com/datasets/rodsaldanha/arketing-campaign

The objective is to enhance your understanding of clustering algorithms and their practical applications in real-world scenarios. Your submission should include a detailed report and a .zip file containing results for the following four sub-questions.

Question 1a

• Import the dataset and prepare the dataset by performing data preprocessing steps, such as handling missing values and scaling numeric features if necessary.
•  Analyze and visualize the dataset to gain insights into its features. This should include exploring distributions, relationships between features, and correlations between features using at least three different visualization plots.

Question 1b

• Design and implement the K-means clustering algorithm in Python for customer segmentation based on their profiles.
• Evaluate the optimal number of clusters using techniques such as the elbow method or silhouette score.
•  Implement K-means clustering with the selected number of clusters.

Question 1c

Appraise the customer segments generated from the clustering process based on their attributes.
Provide insights and recommendations for the company based on the identified customer segments. How can the company leverage these segments for targeted marketing or service improvements?

Question 1d

Create visualizations of the clusters obtained from the clustering algorithm to highlight the key patterns and features.
Evaluate the significant findings, discuss the challenges encountered during the analysis, and outline potential next steps for further improving the clustering analysis.