Inspiration
When we heard the prompt, we racked our brains thinking of a solution, and when one of us saw the Netflix app on our phone, we all thought about how often each of us has opened Netflix to scroll through it for 20 mins and then close it without watching anything. We realized that exploration is not only about exploring outdoors but also exploring your interests and new things.
What it does
This web app takes a person's movie watch history (can be downloaded from streaming services such as Netflix) and computes a dictionary with new movies from the database that are similar to the movies in the person's watch history. Each movie is assigned a sequence of numbers based on their title, genre, and description, and are treated as vectors. The similarities between each movie are computed using the cosine similarity function and assigned a value to be compared altogether.
How we built it
We built the movie recommendation algorithm using Python (with pandas and NumPy) and we built the front end using ReactJS and HTML/CSS. We sent the information from python to the front end using Flask.
Challenges we ran into
Our API worked for the first 70% of the Hackathon but stopped working when we began to integrate the different parts of the program.
Accomplishments that we're proud of
We're proud of the fact that they persevered throughout the HUNDRED of bugs and errors that we encountered. We did not know much about React or even how to use NumPy or pandas and yet were still able to create a completed and finished product.
What we learned
We feel that we learned a lot throughout this experience, particularly using pandas, and scikit as well as Flask and React. I feel that we also learned a lot about data analysis and how to create models to compare data sets.
What's next for Cinematch
Implementing the features that we, unfortunately, did not have the time to complete such as saving the csv file the user inputs onto a database so Flask can easily read it.
Log in or sign up for Devpost to join the conversation.