This is a sandbox system for recommendation engine using the 100K MovieLens Dataset.
It implement the classics recommendation algorithm,such as:
(1) User Based Collaborative Filtering;
(2) Item Based Collaborative Filtering;
(3) Personal Content Based Recommendation Algorithm;
(4) FunkSVD,BiasSVD.
I use several methods to improve the effectiveness of User Based Collaborative Filtering:
(1) Clustering the items using Kmeans and EM;
(2) Using different similarity calculation method,such as Cosine,Pearson Correlation Similarity,Euclidean distance;
(3) Using matrix dimensionality reduction method such as PCA/ICA;
I am adding more and more algorithm and optimization.
- You will need Python version 2.7.5 or up to get started.
- Install scikit-learn. Use scikit-learn Installation Docs.
- Download project files directly using the .tar.gz or .zip links above.
- Hack!
- please waiting..