MLcomp version alpha Statistics: 2045 users, 665 datasets, 389 programs, 18920 runs (0 queued, 4 running), 0 workers

Run 1336

Creator	jake
Program	cberner-KNNImproved-r
Dataset	movielens1m
Task type	CollaborativeFiltering
Created	2y228d ago

Status: Done!

Total Time	1h29m
Max Memory Usage	833M
Domain	CollaborativeFiltering
Learn time
Train RMSE	0
Train MAE	0
Predict train time
Test RMSE	0.905
Test MAE	0.700
Predict test time

Log file

Usage: "run learn trainingFileName" OR "run predict testFileName predictionFileName"
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run learn '/home/mlcomp/worker1/scratch/program0/../dataset2/train'
[1] "starting to build R"
   user  system elapsed 
  1.928   0.132   2.081 
[1] "done building R"
   user  system elapsed 
155.725   0.264 156.565 
[1] "done building B"
[1] "done building S"
[1] 830307
[1] "building new R"
   user  system elapsed 
979.925   2.052 985.875 
[1] "done building new R"
   user  system elapsed 
981.013   2.052 986.972 
[1] "Starting search for best K"
[1] 0.8644662
[1] 0.8642933
[1] 0.8625647
[1] 0.8628665
[1] 0.8626148
[1] 0.8634587
[1] 0.8641642
[1] 0.8647964
[1] 0.8658485
[1] 0.8667533
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run learn '/home/mlcomp/worker1/scratch/program0/../dataset2/train' --- OK [5106s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program3 && ./run stripLabels '/home/mlcomp/worker1/scratch/program0/../dataset2/train' '/home/mlcomp/worker1/scratch/program0/evalTrain.in'
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program3 && ./run stripLabels '/home/mlcomp/worker1/scratch/program0/../dataset2/train' '/home/mlcomp/worker1/scratch/program0/evalTrain.in' --- OK [3s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run predict '/home/mlcomp/worker1/scratch/program0/evalTrain.in' '/home/mlcomp/worker1/scratch/program0/evalTrain.out'
[1] "/home/mlcomp/worker1/scratch/program0/evalTrain.in"
[1] "/home/mlcomp/worker1/scratch/program0/evalTrain.out"
[1] 3.750757
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run predict '/home/mlcomp/worker1/scratch/program0/evalTrain.in' '/home/mlcomp/worker1/scratch/program0/evalTrain.out' --- OK [143s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program4 && ./run evaluate '/home/mlcomp/worker1/scratch/program0/../dataset2/train' '/home/mlcomp/worker1/scratch/program0/evalTrain.out'
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program4 && ./run evaluate '/home/mlcomp/worker1/scratch/program0/../dataset2/train' '/home/mlcomp/worker1/scratch/program0/evalTrain.out' --- OK [6s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program3 && ./run stripLabels '/home/mlcomp/worker1/scratch/program0/../dataset2/test' '/home/mlcomp/worker1/scratch/program0/evalTest.in'
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program3 && ./run stripLabels '/home/mlcomp/worker1/scratch/program0/../dataset2/test' '/home/mlcomp/worker1/scratch/program0/evalTest.in' --- OK [0s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run predict '/home/mlcomp/worker1/scratch/program0/evalTest.in' '/home/mlcomp/worker1/scratch/program0/evalTest.out'
[1] "/home/mlcomp/worker1/scratch/program0/evalTest.in"
[1] "/home/mlcomp/worker1/scratch/program0/evalTest.out"
[1] 3.792601
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program1 && ./run predict '/home/mlcomp/worker1/scratch/program0/evalTest.in' '/home/mlcomp/worker1/scratch/program0/evalTest.out' --- OK [123s]
=== Starting: cd /home/mlcomp/worker1/scratch/program0/../program4 && ./run evaluate '/home/mlcomp/worker1/scratch/program0/../dataset2/test' '/home/mlcomp/worker1/scratch/program0/evalTest.out'
=== Finished: cd /home/mlcomp/worker1/scratch/program0/../program4 && ./run evaluate '/home/mlcomp/worker1/scratch/program0/../dataset2/test' '/home/mlcomp/worker1/scratch/program0/evalTest.out' --- OK [0s]

real	89m41.389s
user	88m39.912s
sys	0m26.570s

Run specification

• supervised-learning: Main entry for supervised learning for training and testing a program on a dataset.
  • (learner:Program) cberner-KNNImproved-r: I implemented KNN with learning to find the best K, using cosine similarity and similarity weighting for prediction. Also the prediction is clamped to the range [1,5] since we know that ratings are in that range.
  • (dataset:Dataset) movielens1m: 1M MovieLens movie ratings dataset from http://www.grouplens.org/. The original dataset contains 1,000,209 anonymous ratings (1-5 stars) of approximately 3,900 movies made by 6,040 MovieLens users who joined MovieLens in 2000. This sub-dataset includes the ratings of 5,000 randomly selected users. A single rating from each user was withheld to form the test set. See included README.txt for more information.
  • (stripper:Program[Strip]) collaborativefiltering-utils: Validates, inspects, and evaluates a dataset in CollaborativeFiltering format.
  • (evaluator:Program[Evaluate]) collaborativefiltering-utils: Validates, inspects, and evaluates a dataset in CollaborativeFiltering format.

Results

• doTest:
  • evaluate:
    • meanAbsoluteError: 0.700145896
    • meanSquaredError: 0.819347882497803
    • numExamples: 5000
    • rootMeanSquaredError: 0.905178370542405
    • success: true
    • time: 0
  • predict:
    • success: true
    • time: 123
  • strip:
    • success: true
    • time: 0
• doTrain:
  • evaluate:
    • meanAbsoluteError: 0.0
    • meanSquaredError: 0.0
    • numExamples: 830307
    • rootMeanSquaredError: 0.0
    • success: true
    • time: 6
  • predict:
    • success: true
    • time: 143
  • strip:
    • success: true
    • time: 3
• exitCode: 0
• learn:
  • success: true
  • time: 5106
• success: true

Comments: