https://github.com/5umitpandey/iit_madras_dataverse
This project presents an end to end data analysis and market intelligence framework built on a global food restaurant dataset.
https://github.com/5umitpandey/iit_madras_dataverse
dataverse eda hackathon iit iitmadras zomato
Last synced: 5 months ago
JSON representation
This project presents an end to end data analysis and market intelligence framework built on a global food restaurant dataset.
- Host: GitHub
- URL: https://github.com/5umitpandey/iit_madras_dataverse
- Owner: 5umitpandey
- Created: 2025-12-17T03:16:09.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2026-01-01T05:14:51.000Z (5 months ago)
- Last Synced: 2026-01-05T23:35:47.767Z (5 months ago)
- Topics: dataverse, eda, hackathon, iit, iitmadras, zomato
- Language: Jupyter Notebook
- Homepage: https://unstop.com/competitions/dataverse-the-business-analytics-challenge-iit-madras-1586925
- Size: 4.77 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# IIT Madras | Dataverse Hackathon
# Food Restaurant Data Intelligence & Market Strategy Analysis
This project presents an end to end data analysis and market intelligence framework built on a global food restaurant dataset.
The objective is to uncover strategic insights across markets, cuisines, pricing, engagement, and growth opportunities using data driven reasoning.
The analysis was developed as part of a hackathon submission and focuses on **business impact**, not just exploratory analysis.
---
## Dataset Overview
The dataset represents a large scale view of restaurant ecosystems across multiple countries and cities.
### Key Statistics
- **Total restaurants**: 9,551
- **Countries covered**: 15
- **Cities covered**: 141
- **Restaurants in India**: 8,652
- **India share**: 90.6 percent
- **Rated restaurants**: 7,403
- **Rating coverage**: 77.5 percent
- **Final feature dimensions**: 22
This immediately establishes India as the dominant market in both scale and engagement.
---
## Strategic Market Positioning
India is the core driver of platform activity.
More than 90 percent of restaurants belong to India, which means:
- Any improvement in ranking, pricing, discovery, or quality directly impacts the majority of users
- India is the most effective market for experimentation and optimization
- Non India markets are better suited for niche and premium strategies
**Conclusion**
India should be treated as the primary experimentation and monetization market.
---
## Data Cleaning & Standardization
Strong insights depend on clean data. The following steps were applied to ensure reliability.
### Country and Currency Fixes
- Numeric country codes were mapped to country names
- Zero missing values after mapping
- Incorrect Philippines currency was fixed
- All costs were standardized to INR for fair comparison
### Column Optimization
Removed non informative columns including:
- Locality Verbose
- Rating color
- Rating text
- Switch to order menu
- Country code
**Business Impact**
Standardized pricing and clean features enable reliable pricing intelligence and cross market analysis.
---
## Cross Country Cost Structure Analysis
Average cost for two was analyzed across countries after INR normalization.
### Market Segmentation
- **India**
- Average cost around 623 INR
- Large volume market
- Best suited for scale driven growth
- **United States and emerging markets**
- Average cost between 2,222 and 2,353 INR
- Medium scale
- Suitable for feature testing
- **UAE and United Kingdom**
- Average cost between 4,113 and 5,785 INR
- Small sample size
- Premium positioning markets
**Insight**
Non India markets have limited samples and should focus on premium and niche strategies rather than scale.
---
## Cuisine Intelligence & Portfolio Strategy
Restaurants serve multiple cuisines stored as comma separated values.
A hybrid feature engineering approach was used to preserve restaurant identity while enabling cuisine level analysis.
### Feature Engineering Flow
1. Original dataset with 9,551 restaurants
2. Controlled cuisine explosion to 19,710 cuisine rows
3. Validation of 145 unique cuisines
---
## Popularity vs Satisfaction Gap
### Most Popular Cuisines
- North Indian: 3,960 restaurants
- Chinese: 2,735 restaurants
- Fast Food: 1,986 restaurants
### Highest Rated Cuisines with scale
- Italian: 3.56 average rating
- Continental: 3.52 average rating
- Cafe: 3.32 average rating
### Popular but Underperforming
- North Indian: 2.51 rating
- Chinese: 2.62 rating
- Fast Food: 2.56 rating
**Key Insight**
Popularity does not imply satisfaction.
Highly popular cuisines require quality standardization, while high rated cuisines should be boosted in discovery systems.
---
## Menu Structure & Engagement Dynamics
Menu diversity plays a critical role in engagement and ratings.
### Cuisine Count Distribution
### Engagement Patterns
- High engagement cuisines consistently exceed 3.5 rating
- Low engagement cuisines cluster below 3.0 rating
**Recommendation**
Recommendation systems should balance popularity with engagement metrics, not popularity alone.
---
## City Level Dynamics in India
### High Volume Cities
**Insight**
This is a discovery problem, not a demand problem.
User engagement exists but ratings are not being captured.
---
## Locality Level Quality Variation
City level strategies are too coarse.
### Example Localities
- Connaught Place
- 3.69 average rating
- Rajouri Garden
- 3.59 average rating
- Shahdara
- 1.41 average rating
**Business Impact**
Locality specific onboarding, audits, and promotions deliver higher ROI than city wide actions.
---
## 📈 Pricing Dynamics & Rating Drivers
### Model Performance
- RMSE: 0.41 on a 5 point scale
- Strong predictive accuracy
### Correlations
- Price range vs rating: 0.44
- Votes vs rating: 0.31
### Feature Importance
- Votes: 71 percent
- Average cost for two: 20 percent
- Online delivery: 4 percent
- Price range: 3 percent
- Table booking: 2 percent
**Strategic Conclusion**
Engagement is the single strongest driver of ratings.
Investment in engagement tools outperforms cosmetic feature additions.
---
### Service Availability Impact on Restaurant Ratings
This section analyzes how **online delivery** and **table booking** features influence restaurant ratings and customer engagement.
The visuals compare rating distributions for restaurants with and without these services, highlighting their relative impact.
| Online Delivery Impact | Table Booking Impact |
|------------------------|----------------------|
|  |  |
---
## Hidden High Quality Growth Opportunities
Certain cuisines show strong ratings with limited competition.
### High Potential Cuisines
| Cuisine | Restaurant Count | Average Rating |
|------------------|------------------|----------------|
| Sandwich | 53 | 4.086038 |
| Steak | 62 | 3.985484 |
| Sushi | 75 | 3.973333 |
| Breakfast | 41 | 3.965854 |
| Mediterranean | 112 | 3.948214 |
| Bar Food | 39 | 3.933333 |
| Indian | 70 | 3.918571 |
| European | 148 | 3.910811 |
| BBQ | 33 | 3.903030 |
| Seafood | 174 | 3.862069 |
**Strategy**
These cuisines are ideal for premium discovery, editorial promotion, and supply expansion.
---
## Final Takeaways
- India is the core growth and experimentation market
- Engagement drives ratings more than pricing or features
- Popular cuisines need quality improvement
- High quality niche cuisines need visibility
- Locality level actions outperform city level strategies
---
## Notebooks and Code
All notebooks used for data cleaning, feature engineering, modeling, and analysis are available here:
🔗 **Drive Link** : https://drive.google.com/drive/folders/1KS_0ilJyk-KP8BpQ8iAdJ--b2_5zZUYC
---
## Team
**Team Name**: ASHSUM
**Team Members**:
This project was built as a hackathon submission with a strong focus on real world business decision making using data.
