MotorbikeRecommenderSys

Inspiration

We chose the MundiMoto challenge because we thought it would be a good idea to make a recommender system to help people find a bike that suits their tastes more easily. Therefore we have made a double recommendation system, one that suggests bikes based on their characteristics (using a content based filtering technique) and another one that suggests bikes based on the interactions of other users (using a collaborative filtering algorithm) . The number of users is 20000 and they have been created randomly.

How we built it

We built it on java with intelij IDEA. We have an executable that takes you to the main menu, where you will find all the motorcycles identified by name and where you can select the user to whom we are recommending. From this view we can navigate through all the motorcycles available in the system. Once selected a motorcycle we can obtain the most similar to this one. (the photo is always the same due to lack of time). From the main menu you can also access to the suggestion obtained with the collaborative filtering algorithm. You can see the readme to see how to execute it.

Challenges we ran into

We wanted to use some API to get the images from the internet, but we couldn't pursue that idea due to lack of time. We also had problems accessing docker.

Accomplishments that we're proud of

We are proud of finishing a project in less than 24 hours, without sleeping.

What we learned

We have learned the techniques to implement collaborative filtering and content based algorithms.

What's next for MotorbikeRecommenderSys

In the future, the project could be expanded with a better graphical interface and a better database to obtain better results. The algorithms used could also be further optimized and their performance improved.