Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/iamantimpal/restorent_data_analysis_and_prediction


https://github.com/iamantimpal/restorent_data_analysis_and_prediction

Last synced: about 2 months ago
JSON representation

Awesome Lists containing this project

README 

        
# Restaurant Insights Analysis and Prediction



**Cognifyz Technologies Verified Project** - 



![MasterHead](https://www.blinkco.io/wp-content/uploads/2021/08/restaurant-data-analytics-illustration.jpg)



Image Courtesy: https://www.blinkco.io/wp-content/uploads/2021/08/restaurant-data-analytics-illustration.jpg



## Problem Statement



**Level 1: Data Exploration and Analysis**



- **Objective:** Gain foundational skills in data exploration, descriptive analysis, and geospatial insights for a restaurant dataset.

- **Tasks:** Explore data structure, address missing values, perform data conversions, and analyze class imbalances. Calculate statistical measures, investigate distributions of categorical variables, and identify top cuisines and cities.

- **Significance:** Develop foundational data science skills for the restaurant industry and data-driven decision-making.



**Level 2: Advanced Analysis**



- **Objective:** Enhance insights from the restaurant dataset through tasks focused on table booking, delivery services, pricing, and feature engineering.

- **Tasks:** Calculate percentages of table booking and online delivery, compare ratings for restaurants with and without booking, and analyze online delivery availability. Determine the most common price range, calculate average ratings for each price range, and associate colors with top-rated ranges. Leverage feature engineering techniques for enhanced data intelligence.

- **Significance:** Deepen understanding and provide valuable insights into customer preferences, services, pricing, and data enrichment.



**Level 3: Predictive Modeling and Insights**



- **Objective:** Perform predictive modeling, customer preference analysis, and data visualization to extract deeper insights.

- **Tasks:** Build regression models to predict restaurant ratings, evaluate model performance, and compare algorithms. Analyze the relationship between cuisine types and ratings, identify popular cuisines, and determine specific cuisines with higher ratings. Create visualizations to depict rating distributions, compare average ratings, and visualize feature relationships.

- **Significance:** Enhance decision-making, understand customer preferences, and predict restaurant ratings, offering comprehensive insights into the restaurant dataset.



---



## Project Summary



**Level 1: Data Exploration and Analysis**



- Conducted comprehensive data exploration and preprocessing to ensure data integrity.

- Performed descriptive analysis, extracting key statistical measures, and identified popular cuisines and cities.

- Explored geospatial insights, visualizing restaurant locations and investigating correlations with ratings.



**Level 2: Advanced Analysis**



- Analyzed table booking and online delivery services, uncovering insights on customer preferences and availability.

- Determined the most common price range and associated it with the highest average rating.

- Employed advanced feature engineering to enhance dataset intelligence.



**Level 3: Predictive Modeling and Insights**



- Built regression models to predict restaurant aggregate ratings, with Random Forest emerging as the top performer.

- Explored the relationship between cuisine types and restaurant ratings.

- Analyzed data visualizations to uncover rating distributions and other insights.



**Overall Insights**



The project covered in-depth data analysis, predictive modeling, and customer preference insights. These findings offer valuable information for data-driven decision-making, customer preferences, and restaurant rating prediction.



---



## Conclusion



**1. Data Overview and Exploration:**



- The dataset encompasses 9,551 restaurant records with 21 columns.

- Minimal null values were detected, predominantly within the 'Cuisines' column.

- No duplicates existed, and data type conversion was unnecessary.

- The 'Aggregate rating' distribution displayed a balanced pattern.



**2. Descriptive Insights:**



- Key statistical metrics for numerical columns were identified.

- The most prominent country codes were 1 and 216, while cities like New Delhi, Gurgaon, and Noida led in restaurant counts.

- Popular cuisines included North Indian and Chinese.



**3. Geospatial Analysis:**



- North America and Asia, especially India, were prominent for restaurant presence.

- New Delhi emerged as the city with the most restaurants, followed by Gurgaon, Noida, and Faridabad.

- Latitude exhibited no correlation with ratings, while longitude showed a negative correlation.



**4. Table Booking and Online Delivery Analysis:**



- Approximately 12.12% of restaurants offered table booking, and 25.66% provided online delivery services.

- Restaurants with table booking displayed a significantly higher average rating of 3.44 compared to 2.56 for those without this service.

- Online delivery was more prevalent in restaurants with medium-priced food products.



**5. Price Range Analysis:**



- Price range 1 was the most common among restaurants.

- Restaurants in price range 4 achieved the highest average rating, followed by price ranges 3, 2, and 1.



**6. Feature Engineering:**



- Two new columns, 'Restaurant Name Length' and 'Address Length,' were created based on the length of restaurant names and addresses.

- Two binary columns, 'Has Table Booking' and 'Has Online Delivery,' were introduced through categorical variable encoding.



**7. Predictive Modeling Insights:**



- Three regression models, Linear Regression, Decision Tree, and Random Forest, were employed to predict restaurant aggregate ratings.

- Random Forest outperformed other models, demonstrating the lowest Mean Squared Error (MSE) and the highest R-squared value.



**8. Customer Preference Analysis:**



- Various cuisines such as cafe, mughlai, north Indian, and fast food had a significant impact on restaurant ratings, with varying performance.

- North Indian and Chinese cuisines displayed greater rating variability, while cafe and fast food cuisines maintained more consistent ratings.

- Based on the number of votes, North Indian, Mughlai, and Chinese cuisines emerged as the most popular.

- Italian, Hawaiian, Seafood, Tea, Sandwich, Continental, and Indian cuisines received the highest average ratings.



**9. Data Visualization Highlights:**



- Restaurant ratings followed a negatively skewed distribution.

- Italian, Hawaiian, Seafood, Tea, Sandwich, Continental, and Indian cuisines secured the top spots in terms of the highest average ratings.

- Cities like Inner City, Quezon City, and Makati City were identified as the most popular based on the highest average rating.

- A positive correlation between votes and restaurant ratings was observed.



This comprehensive project journey provided insights into restaurant data, customer preferences, pricing, services, and predictive modeling, contributing to a holistic understanding of the restaurant industry.



---



## Author



- [Antim Pal](https://www.linkedin.com/in/antim-pal/)



---



## Reference

 - [Cognifyz Technologies Data Science Internship](https://www.cognifyz.com/careers/career.html)