I developed a movie recommender system using a content-based approach for my project. The recommender system analyzes the content and characteristics of movies to provide personalized recommendations to users. The system was implemented using Python programming language and leveraged various tools and libraries.
I started by performing data pre-processing on the movie datasets. This involved cleaning and organizing the data, as well as handling missing values and outliers. The pre-processing steps were carried out in Jupyter Notebook, which provided an interactive and flexible environment for data manipulation.
Afterwards, I moved on to modeling the recommender system using Visual Studio Code (VS Code) and Streamlit. VS Code served as the integrated development environment (IDE) for writing and running the code, while Streamlit was utilized for creating a user-friendly interface to interact with the recommender system.
The content-based approach employed in the system allowed it to recommend movies based on the characteristics and features of previously liked movies by users. The system utilized a dataset comprising approximately 5000 movies, which served as the basis for generating recommendations. Given a specific movie as input, the system would analyze its content and recommend a list of five movies that shared similar characteristics.
The project showcases my proficiency in data pre-processing techniques, Python programming, Jupyter Notebook for data exploration, and using VS Code and Streamlit for developing interactive applications. Additionally, it demonstrates my understanding of content-based recommendation systems and their application in the domain of movie recommendations.
I have used two datasets, and the link to the datasets are: https://www.kaggle.com/datasets/tmdb/tmdb-movie-metadata
The website is made using streamlit and this is the initial design before we click on recommend button to see the five movies just like 'Avatar':
After we click on recommend button, we will see five movies just like avatar based on it's tags:
