Recommender

A recommender system using collaborative filtering (CF) written in python. And compare result between memory-based and low rank matrix factorization approaches.

Requirements

• Python 3+
• Numpy
• Pickle
• sklearn (Only used to caculat root mean squared error(RMSE))

Data

The data is from a collection collected by the GroupLens research group.

Book-Crossing [https://grouplens.org/datasets/book-crossing/]

Preparation

Download the data from GroupLens, uncompress the csv version into /data folder and then run the following scripts

from tool import *
dumpDataMat()

Evaluation

import evaluation
evaluation = Evaluation()
evaluation.evalByAccuracy(recommender = rc.ItemBased(simMeasure=cosSim))
evaluation.evalByAccuracy(recommender = rc.ItemBased(simMeasure=euclidSim))
evaluation.evalByAccuracy(recommender = rc.ItemBased(simMeasure=pearsSim))
evaluation.evalByAccuracy(recommender = rc.UserBased(simMeasure=cosSim))
evaluation.evalByAccuracy(recommender = rc.UserBased(simMeasure=euclidSim))
evaluation.evalByAccuracy(recommender = rc.UserBased(simMeasure=pearsSim))
evaluation.evalByAccuracy(recommender = rc.MatrixFactorization(K = 10))