Exploring cities in India using APIs

IBM Professional Certificate in Data Science Capstone

This project is a part of the IBM Professional Certificate in Data Science Capstone on Coursera.

1. Introduction/Business Problem

The goal of the project is to explore various cities in India to find the city with the best distributions of restaurants, and then explore that particular city to find the types of restaurants, their ratings, and the area with the best restaurants.

The data collected from this study can be used by entrepreneurs dealing with food related businesses or popular food chains looking to establish a new restaurant in India, or even tourists wanting to try out food in India.

2. Data

To provide the necessary information, the following data will be used:
- Cites, Latitude, Longitude: http://www.tageo.com/index-e-in-cities-IN.htm
- Restaurants: Foursquare API, Zomato API

3. Methodology

First, I scraped data regarding Indian cities from http://www.tageo.com/index-e-in-cities-IN.htm using BeutifulSoup and arranged it into a table.

Then, I used Foursquare API to get a list of venues, types, and their location for each city.

After extracting the necessary data, I added weights for each food related venue.

Indian Restaurant:7, Hotel:6, Ice Cream Shop:6, Bar:3, Seafood Restaurant: 5, Coffee Shop:4, Diner: 3, Fast Food Restaurant: 3, Club House:9, Café: 10

Then I calculated the city with the greatest mean weight.

I then used Foursquare API again to get a list of venues within a 100 KM radius of the city Ludhiana, and plotted it on a map using folium.

Now, I shifted my focus to just restaurants. By using the Zomato API, I retrieved restaurant details for the city of Ludhiana.

I began analyzing the retrieved data.

From the plot, we can see that a majority of the ratings are between 3.25 and 4.0. The map below depicts the ratings, with dark green being the best and orange/red being the worst.

Then, I visualized the prices. We can see that a majority of the restaurants have an average price of ₹100– ₹200.

Finally, I arranged the data into two clusters based on price range and rating.

The venues for Cluster 1 have mean price range of 1.00 and rating spread of around 3.56.

The venues for Cluster 2 have mean price range of 1.33 and rating spread around 3.49.

4. Results and Discussion

After our analysis, we selected Ludhiana as our target city, based on the weights applied.

In Ludhiana, we saw that a majority of the venues have a rating of 3.5 and average price per person of ₹100– ₹200.

The results show clustering near Ludhiana junction. There is also not much difference between the two clusters, and the city seems to have evenly spread out restaurants of all types.

5. Conclusion

Therefore, we can conclude that a food enthusiast visiting India might have a good experience in Ludhiana, with good food at decent prices. Even new food shop owners can establish restaurants here as there is demand for all types of food.