Restaurant-Recommendation-System

Using Restaurant Consumer Rating database to predict and recommend top rated restaurants based on user and restaurant profile.

User can also sort the top restaurants accoring to distance based on latitude and longitude values.

Data

The data were originally from the UCI Machine Learning Repository. There are a README and nine csv files in the data directory, including five for the restaurant information, three for the consumer information, and one for the ratings:

Restaurants

• chefmozaccepts.csv
• chefmozcuisine.csv
• chefmozhours4.csv
• chefmozparking.csv
• geoplaces2.csv

Consumers

• usercuisine.csv

• userpayment.csv

• userprofile.csv

Ratings

• rating_final.csv

Three ratings (rating, food rating, and service rating) with values of 0, 1, or 2 are given for a restaurant-consumer pair. More detailed descriptions of the data can be found in the README.

Exploratory Data Analysis(EDA)

Relation of Restaurant Rating with various factors (Price, Parking etc...)

Conclusion: Ratings of restaurant having lower price range are usually lower than those having a meduim/high priced menu

---

Conclusion: Restaurants having a vellet parking options have much higher rating than the restaurants not having one.

---

Conclusion: Restaurants where smoking is allowed have higher overall ratings compared to restaurants not allowing smoking

---

Conclusion: Restaurants having a full bar have better overall ratings than restaurants not serving alcohol or only serving wine and beer

---

Conclusion: Restaurants providing various services including internet had better ratings than restaurats which provided just internet or no services

---

Conclusion: Overall ratings of all restaurants having a formal dress code are much higher than those having informal or no dress code.

---

Conclusions: Restaurants which had closed eating area had higher overall rating than the restaurants which had open eating area.

Accuracy of various models

Accuracy of KNN on K=2 to K=30

Max accuracy at K=5

Hyper-parameter tuning of various models

Accuracy change on changing n_estimators in XG-Boost Classifier Model

---

Accuracy of decision tree classifier on changing tree depth

### Accuracy change on changing min_samples_leaf of decision tree classifier