We noticed that it can be hard to find relatively more obscure games and broaden our horizons for which publishers we considered.
What it does
Recommends video games based on search terms. So you could look for games with gameplay and strategy and our recommender would return the best match it can find in our database with (implied good) gameplay and strategy
How we built it
We used python and also leveraged Microsoft Azure Cognitive Services to do various analytics. Text analytics and word2vec were the two main models used.
Challenges we ran into
We were not very familiar with setting up databases and also had no experience working with azure. So we had to spend a lot of time learning how to work with those services. We also had some scoping issues as for once, we had too much data.
Accomplishments that we're proud of
During testing, we found our model was able to actually able to predict some games very accurately. For example, in a data set of strategy games, when we searched for shooter, we were able to come up with Sanctum 2, a tower defence game with an fps element.
What we learned
This was the first project where we successfully used git for version control. We also learned a lot about data analytics and text analysis. Also go early for bubble tea, they run out quick and M3 locks its door at night.
What's next for Video Game Recommender (AI!!!)
Implement more search functions. Potentially include searching based on video games (aka given a video game, return similar video games, probably not just those from the same series/franchise/publisher), scalability with cloud based databases, more refined scoring algorithm, more efficient database implementation and search functions.