🎬 Movie Recommendation System

This project utilizes data from the TMDb API, containing approximately 10,000 movies, to build a recommendation system.

🔹 Users can select a movie and receive 5 similar recommendations based on machine learning techniques.

➡️ Try it now: Movie Recommendation System

➡️ Giithub Repo: Code link for project

The biggest challenge in this project was fetching data from the TMDb API since it was my first time working with it. Despite the initial difficulty, I gave my best effort and successfully completed the project.

Before this, I attempted a TV show recommendation system, but I wasn’t satisfied with the results. One major issue was the presence of too many news and regular TV shows in the dataset. Removing them significantly reduced the amount of available data, making the project less effective.

Nonetheless, this experience helped me refine my approach, leading to a better movie recommendation system!

Machine Learning Techniques 📐

Stemming:
    - Stemming to reduce words to their base or root form by removing prefixes or suffixes

Bag of Words (BoW):
    - This technique is used to convert textual data into numerical vectors, making it suitable for use in machine learning models.

CountVectorizer: 
        - Text data was transformed into a format suitable for machine learning models using CountVectorizer.  
        - This process represented text numerically, emphasizing the importance of words relative to the dataset.  

Cosine Similarity:
    - Cosine Similarity is a measure used to determine how similar two text documents (or vectors) are, by calculating the cosine of the angle between them.

Saving the train model
    - I trained the file using cosine similarity and saved it. The file was about 550 MB, making it difficult to upload to GitHub. So, I uploaded it to Google Drive and reloaded it from there whenever I wanted to deploy the model on Streamlit. It takes a while depending on the internet speed, but once completed, it works perfectly fine.

🛠️ Tools & Software