🛍️ Purchase Prediction Engine

A machine learning project that predicts customer purchase amounts using demographic, geographic, and product-related features.

The project uses Linear Regression to model the relationship between customer and product attributes and purchase amount, followed by evaluation on unseen test data.

---

📌 Project Overview

Understanding customer purchasing behaviour is an important problem in retail analytics. Transactional datasets contain valuable information about customers, products, demographics, and purchasing patterns that can be used to build predictive models.

The goal of this project is to develop a regression-based machine learning pipeline that can:

• Predict the purchase amount for a customer-product combination
• Analyse how customer and product attributes relate to purchasing behaviour
• Evaluate prediction performance using standard regression metrics
• Generate purchase predictions for unseen customer records

🎯 Prediction Target

Input: Customer and product attributes

Output: Predicted Purchase amount

---

📊 Dataset

The project uses a retail purchase dataset containing 550,068 transactions and 12 attributes.

Feature	Description
User_ID	Unique customer identifier
Product_ID	Unique product identifier
Gender	Customer gender
Age	Customer age group
Occupation	Customer occupation category
City_Category	Category of the customer's city
Stay_In_Current_City_Years	Number of years living in the current city
Marital_Status	Customer marital status
Product_Category_1	Primary product category
Product_Category_2	Secondary product category
Product_Category_3	Tertiary product category
Purchase	Purchase amount - prediction target

The observed purchase amounts range from 12 to 23,961, with an average purchase amount of approximately 9,264.

---

🔄 Machine Learning Pipeline

The project follows a typical supervised machine learning workflow:

Raw Dataset
     │
     ▼
Data Exploration
     │
     ▼
Data Preprocessing
     │
     ├── Missing Value Handling
     ├── Categorical Encoding
     ├── Feature Transformation
     └── Feature Scaling
     │
     ▼
Train / Test Split
     │
     ▼
Linear Regression
     │
     ▼
Model Evaluation
     │
     ├── MAE
     ├── RMSE
     └── R² Score
     │
     ▼
Purchase Prediction

---

🤖 Machine Learning Model

Linear Regression

Linear Regression is used as the primary regression model to predict purchase amounts from customer and product attributes.

The modelling pipeline includes:

• Categorical feature encoding
• Numerical feature transformation
• Feature scaling
• Linear Regression
• Model evaluation on unseen test data

The model is implemented using scikit-learn.

---

📈 Model Performance

The model was evaluated using three standard regression metrics:

Metric	Result
MAE	2,208.91
RMSE	2,946.09
R² Score	0.6546

The model achieved an R² score of 0.6546 on the test set, indicating that approximately 65.46% of the variation in purchase amounts is explained by the features used in the model.

Train vs Test Performance

Dataset	R² Score
Training	0.6619
Testing	0.6546

The relatively small difference between the training and testing R² scores indicates that the model performs similarly on both datasets.

---

📉 Prediction Results

The following visualization compares the actual purchase amounts with the predicted purchase amounts generated by the Linear Regression model.

---

🧰 Technologies Used

Technology	Purpose
Python	Core programming language
Pandas	Data manipulation and analysis
NumPy	Numerical computation
Matplotlib	Data visualization
Seaborn	Exploratory data visualization
Scikit-learn	Machine learning and model evaluation
Jupyter Notebook	Experimentation and analysis

---

▶️ Running the Project

1. Clone the repository

git clone https://github.com/simran2104/purchase-prediction-engine.git
cd purchase-prediction-engine

2. Install dependencies

pip install -r requirements.txt

3. Run the notebook

Open:

Purchase-Prediction.ipynb

Execute the notebook cells to:

• Explore the dataset
• Perform data preprocessing
• Train the regression model
• Evaluate model performance
• Visualize predictions
• Generate purchase predictions