User Based Movie Recommendation Engine Using Netflix Movie Dataset

Domain : Artificial Intelligence, Machine Learning.

Sub-Domain : Recommendation Engine, Recommender System.

Techniques : Collaborative Filtering, User-User Collaborative Filtering.

Description

1. Developed user-based movie recommendation system by implementing user-user collaborative filtering.
2. Used Netflix movie dataset containing 100,000 user records for developing recommendation engine.
3. Reduced run time and space complexity significantly.
4. Implementation in both C++ and Python separately.

For Python implementation:

1. Please open "AnjanaTihaMachineLearningProjectvFinal.ipynb" file in python notebook(Anaconda contains most libraries)
2. Provide "movie_titles.txt" and "ratings.txt" file location in each file read function for moie_titles and ratings.
3. Run all cells in the file.
4. After running, user id and recommendation size K will be asked.
5. Upon providing user id and recommendation size, K titles and year of recommended movies will be displayed from most most recommended in descending order.

For C++ implementation:

1. Please open "movie_recom.cpp" file in a isual Studio/Eclipse/ other C++ software tools project. Use an ecplipse project already provided under the zip file(Contains two seperate implementation for C++ and Python). Please use 7zip software for unzipping "http://www.7-zip.org/download.html"
2. Provide "movie_titles.txt" and "ratings.txt" file location in argument in Visual Studio/Eclipse/ other C++ software tools in source code.
3. Build and Run.
4. After running, user id and recommendation size K will be asked.
5. Upon providing user id and recommendation size, K titles and year of recommended movies will be displayed from most most recommended in descending order. Note: C++ implementation used C++ 11 version.

Languages : Python, C++

Tools/IDE : Anaconda, Visual Studio

Libraries :

Duration : February - April 2017

Current Version : v1.0.0.0

Last Update : 04.30.2017