Bookstore Customer Analytics:
Predicting Customer Spending and eBook Subscription Behavior
Project Overview
For this project, I explored a bookstore customer dataset to understand purchasing behavior and identify customers who may be interested in an eBook subscription service.
The project combined exploratory data analysis, machine learning, and dashboard development to transform customer data into actionable business insights.
Business Goals
The project focused on two key objectives.
1. Monthly Spend Prediction
Develop a regression model to estimate each customer's average monthly spending.
2. eBook Subscription Prediction
Build a classification model to identify customers likely to subscribe to the bookstore's eBook service.
Tools Used
Programming and Analysis
- Python
- Pandas
- YData Profiling
Visualization
- Tableau
- AutoViz
Machine Learning
- AutoGluon
- Scikit-learn
Development Environment
- Google Colab
Data Preparation
The dataset required several preprocessing steps before model development:
- Removed unnecessary identifiers
- Handled missing values
- Cleaned categorical variables
- Explored feature distributions
This stage reinforced the importance of data quality before building predictive models.
Exploratory Data Analysis
Several patterns emerged during exploration:
- Income levels influenced customer spending behavior
- Certain Occupation group showed higher purchasing activity
- Subscription rates varied across customer segments
- Spending patterns differed among income categories.
Visualization helped uncover relationships that were not immediately visible in the raw data.
Machine Learning Approach
Regression Model
Goal:
Predict average monthly spending
Skilled Practiced:
- Feature selection
- Model training
- Regression evaluation
- Mean Squared Error(MSE)
- Root Mean Squared Error(RMSE)
Classification Model
Goal:
Predict likelihood of eBook subscription
Skilled Practiced:
- Binary classification
- Probability prediction
- Model evaluation
- Feature importance analysis
AutoGluon simplified model experimentation and allowed comparision of multiple algorithms.
Dashboard Development
After completing the analysis, I created Tableau dashboards to communicate findings.
Dashboard sections included:
- Customer Demographics
- Spending Patterns
- Subscription Analysis
- High-Potential Customer Segments
The goal was to present insights in a format that business stakeholders could easily understand.
Key Skilled Gained
Through this project, I strengthened my abilities in:
- Data Cleaning
- Exploratory Data Analysis
- Feature Engineering
- Regression Modeling
- Classification Modeling
- Tableau Dashboard Design
- Business Storytelling
- Data Visualization
- Machine Learning Workflow
Lesson Learned
One of the biggest lessons from this project was that successful machine learning depends heavily on understanding the data and the business problem.
Building accurate models is important, but communicating insights effectively through dashboards and storytelling is equally valuable.
Next Steps
Future improvements could include:
- Advanced customer segmentation
- Customer lifetime value prediction
- Recommendation systems
- Marketing campaign optimization
- Time-series sales forecasting
This project was an excellent opportunity to practice end-to-end data science, from raw data to business insights.