This project aims to analyze customer behavior in response to promotional activities conducted by Starbucks through its app. By leveraging transaction data, demographic information, and records of push notifications sent to users, we seek to predict customer responsiveness to promotions such as advertisements, discount coupons, or Buy One Get One (BOGO) offers.
The dataset consists of three main components:
- portfolio.csv: Contains the IDs and metadata (duration, type, etc.) of each promotional offer.
- profile.csv: Demographic data of each customer.
- transcript.csv: Records of transactions, received offers, viewed offers, and completed offers.
The project comprises three main code files:
-
Exploratory Data Analysis (EDA) (
eda.py): This file contains code for initial data exploration. It includes data cleaning, basic statistical analysis, and visualization of the datasets to uncover patterns, anomalies, and relationships. -
Customer Clustering (
clustering.py): This script applies unsupervised learning techniques to understand the customer base. It segments customers into distinct groups based on their transaction and demographic data, providing insights into different customer profiles. -
Promotion Response Classification (
classification.py): The final code file focuses on predicting whether a customer will complete a promotional offer. It uses classification algorithms to model the likelihood of offer completion based on customer characteristics and historical response data.
To run these scripts, you will need Python installed on your machine along with the following libraries:
- Pandas
- NumPy
- Matplotlib
- Scikit-learn
Install these libraries using pip:
To execute the scripts, run the following commands:
- python eda.py
- python clustering.py
- python classification.py
Contributions to this project are welcome. Please read CONTRIBUTING.md for details on our code of conduct and the process for submitting pull requests.
This project is licensed under the MIT License - see the LICENSE.md file for details.
- Starbucks for providing the dataset.
- The machine learning community for valuable insights and tools.