Average MAE over 25 randomAL runs:
0.84 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.82 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.83 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 |
Problem: The numbers are not decreasing much (contradictory to expectations)
ToDo: Need to find the problem - check the code
Log:
It is not caused by the old bug in GenericUserBasedRecommender
Correlation
The problem might lie in the correlation function. When using PearsonCorrelation get very bad results MAE = max; when using UnceteredPearson Correlation get much better results MAE=0.9.
There was a small bug in the code (submitted to the SourceForge).