The Car Recommendation System recommends cars to a user based on certain criteria that they input. The car data is retrieved from a database and read into a vector data structure. Next, user input for certain criteria is requested. Then, the program searches through the vector for the most relevant cars. Finally, the relevant cars are outputted to the user
Created by Amay Patel, Rohan Chander, and Saurabh Anand for COP3530 - Data Structures & Algorithms on 08/01/2022
The hierarchy of criteria that guides our insertion and searching algorithm for the B+ Tree is as follows (in other words, these are the parameters the user will be searching for in order of importance):
- Price tag (Range)
- Brand (Select values)
- Year* (Range)
- City MPG (Range)
- Highway MPG (Range)
- Fuel type* (Select values)
- Horsepower (Range)
*Required
The user can enter multiple values for a parameter or type "None" which searches for all values of the parameter. Once a list of cars that meet the user's criteria is complete, we will compute the cosine similarity on each car with the user's ideal car (the ideal car is created by averaging the values specified in price tag, city mpg, highway mpg, and horsepower). The 50 cars with the highest similarity will be returned with the first car in the output being the most similar to the user's preferences. If there are less than 50 cars that match the users preferences, we will return all of the cars with the first car being the car that is most similar to the user's preferences.
When the user is done inputting their preferences for the car recommendation system, they will be asked whether they would like to use the MergeSort or TimSort algorithm for their recommended cars list.
MergeSort is a divide-and-conquer sorting algorithm used to sort the vector in our program. The divide step divides the vector into multiple small vectors while the conquer step merges the small vectors a bigger and sorted vector. It has best, average, and worst case time complexity of O(nlogn).
TimSort is a unique sorting algorithm that is derived from the InsertionSort and MergeSort algorithms. It has a best case time complexity of O(n) and an average and worst case time complexity of O(nlogn).