Inspiration
Sounded cool and approachable from a mathematical standpoint.
What it does
It is a jupyter notebook which creates a graph of all products, with edges connected by customers traversing the graph. Recommendations are created based off the weights of the edges adjacent to the path, which depends on the time since the product was viewed (using an exponential decay).
How we built it
We had an initial idea of looking at neighbours of the path, and we stuck with this. We realised that its a good idea to also incorporate the top liked items by everyone as well.
Challenges we ran into
A lot of our time went into running the program again and again, and we lost a lot of time on code we ultimately didn't need, as we realised how to greatly simplify our work.
Accomplishments that we're proud of
We got a working model.
What we learned
Brainstorm and have a much more detailed plan before writing up the code.
What's next for Recommendation Engine
Possible AI implementation.
Log in or sign up for Devpost to join the conversation.