b1sakher / POC-New-scoring-model-for-ranking-cars-of-Drivy-search-Engine

The goal is to make a proof of concept of a new ranking algorithm for Drivy's search engine.

POC-New-scoring-model-for-ranking-cars-of-Drivy-search-Engine

The goal is to make a proof of concept of a new ranking algorithm for Drivy's search engine.

The goal of this test is to make a proof of concept of a new ranking algorithm for Drivy's search engine.

How does the search engine work?

Users input an address of search and desired dates and time for their trip. (screenshot)

Optionally, users can filter the results on various criteria such as the type of car (city, sedan, commercial vans..), price, options (AC, snow tires..) etc..

The search engine renders a list of cars. It follows 2 main steps to do so:

Filtering: Generate the list of cars that matches the search request: ie. Cars nearby the address, that are available for the dates of search, and that match the selected filters. This is not in the scope of this project.

Ranking: Order the list based on a score. Current score is computed with a simple heuristic. The scope of this project is precisely this ranking part.

Anytime the user changes a parameter, filter or page number, a new request is sent to the search engine and the results refresh.

Users can then visit different car listings, select cars and eventually book one of them.