1. Python 3.6 and libraries

• Numpy
• Matplotlib
• scikit-learn

2. Anaconda

• Jupyther Notebook
• PyTorch

All copyrights for the data is owned by Zomato Media Pvt. Ltd.. data is taken from here and is only for educational purpose

Thank you Himanshu Poddar for providing the data set in kaggle for others to use it for educational purpose https://www.kaggle.com/himanshupoddar/zomato-bangalore-restaurants

The overall rating on Zomato for a restaurant is not the average rating given by the customers. The rating is relative to other restaurants in the same city. Zomato has many algorithms to determine the final rating which can change with performance of other restaurants near by as well. The final ratings all the restaurants of a city is placed on a normal curve. Which makes them relative to each other. .(Zomato rating system) Some known factors are:

• Frequency of ratings/reviews by customers
• Rates/reviews of other nearest restaurants Other factors are:
• Location of the restaurant
• Cuisines served
• Demography of the location
• Services of the restaurant – online service, cuisines
• theme of the restaurant - Café, restaurant, pub, bar
• category of the food served. The final rating for the restaurant is in between 1 – 5 with interval of .1. The ratings can be like 1,1.1,1.2,…,4.8,4.9 or 5 for the overall rating.

Being an IT professional my self I have been dependent on food applications to search for meals when I chose not to cook. Prices are very important for me and to check if the food quality is good I find myself quickly scanning through the ratings based on which I think if the restaurants use good oil, prepares food properly, hygienic or not. Choosing from the same restaurants can be boring which pushes me to look for other restaurants for different taste and food. While choosing a new restaurant I look for rating before actually looking at the Menu.

zomato_bangalore_report2.ipynb - Jupyter notebook containing the systematic process for data exploration, pre-processing, training and predicting

train.py - Python code to train a neural network
predict.py - Python code to process and predict input features for prediction
helper.py - helper fucntions for pre-processing the data
visuals.py - functions to visualize the columns of the restaurnt dataset
rates.json - json file to convert label to category/class
report.pdf - Report explaining the project in detail
images- images used in notbook report

Image shows 3 top rating bins with probability of predicting these class