PickFlicks

• 

  Recommending for The Dark Knight Rises

• 

  Recommending for Avatar

Inspiration

While looking for what to watch on Netflix I get bombarded with suggestions which don't really help me. I wanna watch something in my genre, from my favourite director, or even having my fav actors. Enter PickFlicks.

What it does

PickFlicks uses content filtering algorithm to group similar movies together based on genres, description, actor, director, title etc.

How we built it

I trained a model to find likeness between movies from the TMDB dataset available on Kaggle. Then I created a similarity matrix based on this data. Finally I dumped this data and created a website using Streamlit for the user to get movie recommendations from.

Challenges we ran into

Lots of problems managing the dataset, cleaning the data, creating a recommendation function.

Using the TMDB API for posters was very challenging at first and a new experience to me.

Deploying and Hosting on Streamlit was a whole different level of headache in itself.

Accomplishments that we're proud of

The recommendation system works, and provides movies very well

What we learned

ML, cosine similarity, NLP, Streamlit, Python

What's next for PickFlicks

A Netflix like hover preview instead of static posters

Built With

• apis
• kaggle
• natural-language-processing
• python
• scikit-learn