SuperKart is an AI-driven retail analytics and customer segmentation platform designed to enhance business decision-making in the retail and e-commerce sectors.
It uses Machine Learning (ML) and Predictive Analytics to provide insights into customer purchasing behavior, optimize marketing strategies, and increase profitability through data-driven recommendations.
- Identify key customer segments and their purchasing patterns.
- Predict customer churn and purchase likelihood using ML models.
- Enable targeted marketing campaigns to improve engagement and retention.
- Support data-driven inventory and pricing decisions.
Retailers often struggle to:
- Understand evolving customer behavior.
- Personalize marketing at scale.
- Optimize inventory and promotional strategies.
SuperKart addresses these challenges by providing an AI-powered analytics system capable of predicting sales trends, segmenting customers, and identifying churn risks—leading to higher ROI and customer retention.
| Attribute | Description |
|---|---|
| Dataset | SuperKart Customer & Transaction Data |
| Records | 12,345 customers |
| Features | Customer demographics, purchase history, spending score, product category, feedback |
| Target Variable | Customer churn / Purchase prediction |
- Removed missing and duplicate entries.
- Encoded categorical features (gender, region, etc.).
- Normalized numerical attributes (age, income, purchase frequency).
- Handled class imbalance using SMOTE.
- Split data into Train (70%), Validation (15%), and Test (15%) sets.
| Model | Purpose | Key Features | Result |
|---|---|---|---|
| Logistic Regression | Baseline classifier | Simple interpretability | Accuracy: 0.78 |
| Random Forest | Churn prediction | Ensemble learning | Accuracy: 0.89 |
| XGBoost | Purchase likelihood | Gradient boosting | Accuracy: 0.91 |
| K-Means | Customer segmentation | Behavioral clustering | 4 meaningful clusters |
- High-income, high-frequency buyers represent the most loyal segment.
- Customers with declining purchase frequency show early signs of churn.
- Regional and seasonal variations strongly affect purchasing behavior.
- Personalized offers increase repeat purchases by 22%.
| Metric | Logistic Regression | Random Forest | XGBoost |
|---|---|---|---|
| Accuracy | 0.78 | 0.89 | 0.91 |
| Precision | 0.80 | 0.90 | 0.93 |
| Recall | 0.76 | 0.87 | 0.92 |
| F1 Score | 0.78 | 0.88 | 0.92 |
XGBoost emerged as the best-performing model and was selected for deployment.
| Cluster | Description | % of Customers | Strategy |
|---|---|---|---|
| 0 | High-value loyal customers | 25% | Reward with loyalty perks |
| 1 | Low-frequency price-sensitive | 30% | Target with discounts |
| 2 | New/occasional buyers | 20% | Promote engagement campaigns |
| 3 | High-churn-risk customers | 25% | Offer personalized retention offers |
- 📊 Improved customer retention through churn prediction (accuracy 91%)
- 🧠 Personalized marketing boosted engagement by 22%
- 🏷 Dynamic pricing and stock optimization based on predictive trends
- 💡 Actionable insights for management via intuitive dashboards
- Integrate real-time recommendation engine for online customers
- Add deep learning models (LSTMs) for time-series sales forecasting
- Include NLP-based sentiment analysis on customer feedback
- Expand to multi-channel data sources (web, mobile, in-store)
📦 superkart-analytics/ ├── data/ # Raw and processed data files ├── notebooks/ # Jupyter notebooks for EDA and modeling ├── src/ │ ├── preprocessing.py │ ├── model_training.py │ ├── segmentation.py │ └── evaluation.py ├── models/ # Trained models (pickle files) ├── reports/ # Results and visualizations ├── requirements.txt └── README.md