likicode/BigDataMining-Analysis

movie recommendation

Project for Big Data Mining and Analysis Course

Movie Rating Prediction Project

• Dataset

  • Download from http://files.grouplens.org/datasets/ movielens/ml-1m.zip.
  • The data set contains 1 million ratings from 6000 users on 4000 movies.
  • We further sort each ratings by timestamp

• Recommendations:

  • Step 1:
    • Baseline estimator: use the formular bxi = μ + bx + bi on pdf
  • Step 2:
    • Neighborhood estimator: use the neighborhood approach to predict rating score
      • item-based similarity
      • user-based similarity
  • Incorporating Temporal Dynamics
    • Following the KDD09 Paper http://dl.acm.org/citation.cfm?doid=1721654.1721677
    • Equation 5,6,8,10

• K-mean Clustering

  • Use the k-mean algorithm to cluster the users based on their rating scores given by the file ratings.dat.

• SVD Dimensionality Reduction

  • Use the SVD algorithm to reduce the dimensionality

• Metrics

  • the value of RMSE

• Project Implementation

  • Please refer to https://github.com/likicode/datamining_course_proj/blob/master/数据挖掘文档.pdf
  • Recommendations part is coded with Python, the others are coded with Matlab by my group member.