⚡ QuickMind — AI-Powered Quick Commerce Delivery Intelligence System

End-to-End Analytics & ML Prediction Platform for Quick Commerce Delivery Operations
Modeled after Blinkit · Zepto · Swiggy Instamart · Amazon Now

---

🚀 Live Demo

🌐 Live App → https://quickmind-analyticsbranchmainmainfileapppy-3hbabnjkru4eprjiso6.streamlit.app

Login: admin / admin123

---

Run Locally

streamlit run app.py
→  http://localhost:8501

---

📋 Project Overview

QuickMind is a production-grade data science project that combines real-world analytics, business intelligence, and machine learning to solve operational challenges in the quick commerce (10-minute delivery) industry.

Attribute	Detail
Dataset Size	9,36,453 rows × 13 columns
Companies	Blinkit, Zepto, Swiggy Instamart, Dunzo + 4
Cities	12 major Indian metros
ML Model	Random Forest (R²=0.9658, MAE=0.95 min)
Stack	Python, Streamlit, Plotly, Scikit-Learn, XGB
Deployment	Streamlit Cloud / Local

---

🎯 Business Problem Statement

Quick commerce platforms promise delivery under 30 minutes — a critical differentiator. Yet SLA breaches are frequent due to:

• Variable delivery distances
• Partner performance inconsistency
• Demand spikes in specific cities
• Product category complexity

Solution: An end-to-end intelligence system that:

1. Monitors SLA compliance in real-time
2. Predicts delivery time using ML (R² = 0.9658)
3. Identifies root causes of breaches
4. Generates automated business recommendations

---

🏗️ Architecture

Raw Data (CSV)
     ↓
Data Cleaning (Jupyter/VS Code)
     ↓
Cleaned Dataset ──→ EDA (eda_analysis.py)
     ↓                      ↓
Feature Engineering    Charts & Insights
     ↓
ML Pipeline (train_model.py)
  ├── Linear Regression  (baseline)
  ├── Random Forest      ✅ BEST (R²=0.97)
  └── XGBoost            (comparison)
     ↓
Saved Models (.pkl)
     ↓
Streamlit App (app.py)
  ├── Admin Login
  ├── Executive Dashboard
  ├── Analytics & EDA
  ├── Operations Dashboard
  ├── AI Predictions
  ├── Business Insights
  ├── Data Explorer
  └── Upload & Refresh

---

📁 Project Structure

quickcommerce/
│
├── 📂 data/
│   └── cleaned_quick_commerce.csv       # Cleaned dataset (9.4L rows)
│
├── 📂 models/
│   ├── best_model.pkl                   # Best ML model (Random Forest)
│   ├── random_forest.pkl                # Random Forest model
│   ├── xgboost.pkl                      # XGBoost model
│   ├── linear_regression.pkl            # Linear Regression baseline
│   ├── label_encoders.pkl               # Categorical encoders
│   ├── feature_names.pkl                # Feature list
│   └── model_metrics.csv                # Comparison metrics
│
├── 📂 sql/
│   └── queries.sql                      # 30+ SQL analytics queries
│
├── 📂 notebooks/
│   └── *.png                            # EDA charts (auto-generated)
│
├── 📂 assets/                           # Static assets
│
├── app.py                               # 🌐 Main Streamlit application
├── train_model.py                       # 🤖 ML training pipeline
├── eda_analysis.py                      # 📊 EDA script (generates charts)
├── requirements.txt                     # 📦 Python dependencies
└── README.md                            # 📖 This file

---

📊 Dataset Columns

Column	Type	Description
Order_ID	int	Unique order identifier
Company	str	Delivery platform name
City	str	Delivery city
Customer_Age	int	Age of customer (18–59)
Order_Value	int	Order value in INR
Delivery_Time ✅	int	Minutes taken to deliver (TARGET)
Distance_km	float	Delivery distance in km
Items_Count	int	Number of items in order
Product_Category	str	Category of products ordered
Payment_Method	str	Payment mode used
Customer_Rating	int	Rating given by customer (1–5)
Discount_Applied	int	Whether discount was applied (0/1)
Delivery_Partner_Rating	int	Delivery partner's rating (2–5)

---

🤖 Machine Learning Results

Model	MAE (min)	RMSE (min)	R² Score
Linear Regression	4.0765	5.3361	0.2162
Random Forest ✅	0.9548	1.1144	0.9658
XGBoost	0.9554	1.1016	0.9666

Winner: Random Forest selected as best model by MAE.

Feature Importance (Top 5)

1. Distance_km — strongest predictor
2. Delivery_Partner_Rating — operator quality signal
3. Items_Count — order complexity
4. Order_Value — basket size correlation
5. Company — platform-specific logistics

---

📌 KPIs & Business Metrics

KPI	Value	Business Meaning
SLA Breach Rate	~X%	% of orders beyond 30-min target
Avg Delivery Time	~X min	Core performance indicator
Avg Customer Rating	~3.5 / 5	Customer satisfaction health
Revenue per Order	~₹X	Monetization efficiency
Partner Rating	~3.8 / 5	Delivery fleet quality
Discount Rate	~40%	Promotion strategy indicator

---

⚙️ How to Run

1. Install dependencies

pip install -r requirements.txt

2. Train ML models

python train_model.py

3. (Optional) Generate EDA charts

python eda_analysis.py

4. Launch the dashboard

streamlit run app.py

Open: http://localhost:8501
Login: admin / admin123

---

☁️ Deploy on Streamlit Cloud

1. Push this project to GitHub
2. Go to share.streamlit.io
3. Connect your repo → select app.py
4. Add data/ folder with the CSV
5. Click Deploy → share the URL!

Note: Run python train_model.py locally first, then commit models/ to GitHub before deploying so the pre-trained models are available in the cloud.

---

🔑 Admin Login

Username	Password	Role
admin	admin123	Admin
analyst	analyst123	Analyst
viewer	viewer123	Viewer

---

💡 Key Business Insights

1. SLA Breach Pattern: Long-distance orders (>15 km) have 3× higher breach rates
2. Partner Quality: 5★ partners deliver 40% faster than 2★ partners
3. Discount Impact: Discounted orders average ~₹50 higher value
4. City Variance: Metro cities show 15-25% SLA breach vs 5-10% in smaller cities
5. Category Speed: Dairy & Fresh items have fastest delivery; Personal Care is slowest
6. Payment Insight: UPI/Wallet users order ~8% more frequently than COD users

---

🔮 Future Scope

• Real-time data streaming via Kafka/Pub-Sub
• Time-series forecasting (Prophet/LSTM) for demand prediction
• Route optimization API integration (Google Maps/Mapbox)
• Deep learning model (Tabular DNN) for higher accuracy
• Multi-language dashboard support
• Mobile-responsive PWA version
• Slack/email SLA breach alerting
• A/B test dashboard for discount strategy

---

📄 Resume Description

AI-Powered Quick Commerce Delivery Intelligence System (Final Year Project)
Built an end-to-end analytics and prediction platform for quick commerce delivery operations using Python, Streamlit, and machine learning. Analyzed 9.4 lakh orders across 8 platforms and 12 Indian cities. Developed a Random Forest model achieving R²=0.9658 for delivery time prediction. Created an interactive 7-page Streamlit dashboard with admin authentication, real-time KPIs, EDA charts (Plotly), SLA breach monitoring, and AI-powered predictions. Generated 30+ SQL queries for business intelligence. Delivered auto-generated business insights with actionable recommendations.
Tech Stack: Python, Pandas, NumPy, Scikit-Learn, XGBoost, Streamlit, Plotly, Matplotlib, Seaborn, SQL

---

🛠️ Tech Stack

Frontend:    Streamlit + Plotly + Custom CSS (Google Fonts)
Backend:     Python 3.10+
ML:          Scikit-Learn, XGBoost, Joblib
Data:        Pandas, NumPy
Viz:         Plotly Express, Plotly Graph Objects, Matplotlib, Seaborn
Database:    PostgreSQL (SQL queries provided)
Deployment:  Streamlit Cloud / Docker

---

Developed as part of a Data Science Internship project focusing on real-world quick commerce analytics. This system demonstrates production-grade engineering across the full data science lifecycle — from raw data ingestion and cleaning, through exploratory analysis and machine learning, to interactive dashboard deployment. Built to solve real operational challenges faced by quick commerce platforms like Blinkit, Zepto, and Swiggy Instamart in the Indian market.