An exploratory data science project that turns raw e-commerce transaction data into an interactive Streamlit dashboard.
project-foresight-uk4v.onrender.com
⚠️ Hosted on Render's free tier — the app sleeps after inactivity, so the first load after idle time can take 30–60 seconds to spin up. If you're sharing this link with a recruiter, mention that upfront so a slow first load doesn't read as broken.
Project FORESIGHT analyzes retail transaction data through a series of Jupyter notebooks — cleaning, EDA, feature engineering, customer segmentation, demand forecasting, and inventory recommendation — and presents the output through a Streamlit dashboard deployed on Render.
This is the actual structure — nothing here is invented:
Project_FORESIGHT/
├── app/
│ └── app.py # Streamlit dashboard entry point
├── data/ # Raw and processed datasets
├── models/ # (currently empty — no serialized models saved yet)
├── notebooks/
│ ├── 01_Data_Cleaning.ipynb
│ ├── 02_EDA.ipynb
│ ├── 03_Feature_Engineering.ipynb
│ ├── 04_Customer_Segmentation.ipynb
│ ├── 05_Demand_Forecasting.ipynb
│ ├── 06_Inventory_Recommendation.ipynb
│ └── 07_Dashboard_Preparation.ipynb
├── reports/ # (currently empty)
├── screenshots/
│ ├── dashboard.png
│ ├── analytics.png
│ ├── map.png
│ └── insights.png
├── src/ # (currently empty — pipeline not yet modularized)
├── requirements.txt
└── README.md
01_Data_Cleaning → handle missing IDs, cancellations, duplicates
02_EDA → distribution, trend, and correlation analysis
03_Feature_Engineering → RFM and time-based features
04_Customer_Segmentation→ K-Means clustering on RFM features
05_Demand_Forecasting → category/SKU-level demand forecasting
06_Inventory_Recommendation → reorder signals from forecast output
07_Dashboard_Preparation→ aggregation for the Streamlit app
│
▼
app/app.py (Streamlit) → deployed on Render
Screenshots below are the actual files from screenshots/ —
Executive Overview
Analytics View
Geographic Revenue
Business Insights
| Layer | Technology |
|---|---|
| Language | Python 3.10+ |
| Data processing | Pandas, NumPy |
| Machine learning | Scikit-learn (K-Means) |
| Visualization | Plotly |
| Dashboard | Streamlit |
| Analysis environment | Jupyter Notebook |
| Deployment | Render |
git clone https://github.com/Yashr4635/Project_FORESIGHT.git
cd Project_FORESIGHT
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
streamlit run app/app.pyOpen http://localhost:8501 in your browser.
Being direct about project status, because vague completeness claims fall apart under questioning:
Done:
- Data cleaning notebook
- EDA notebook
- Feature engineering (RFM)
- Streamlit dashboard app connected to processed data
- Customer segmentation via K-Means (notebook-based)
- Demand forecasting notebook
- Inventory recommendation notebook
- Live deployment on Render
D.S. Yashaswi GitHub
MIT