Benchmarks

Scikits.learn benchmark

This is from a benchmark developed by the scikits.learn team. I ran it on my Intel Core2 6600, 2.40GHz CPU.

Results in scikits.learn ml-benchmarks
Benchmark PyMVPA Shogun Pybrain MLPy scikits.learn milk
knn 1.0 2.23 2.23 3.05 2.20
elasticnet 174.43 1.0
lassolars 61.67 1.0
pca 1.0 11.11
kmeans 2.02 7057.02 1.61 6.74 1.0
svm 3.35 1.20 1.24 1.0

All of the results are normalised by the fastest system for each entry (which is therefore, by definition, 1.0).

So, except for PCA, milk is pretty fast and for kmeans and SVM learning it is the fastest system.

Limitations of This Benchmark

  1. It is very small dataset, so you do not get a feeling of how it scales.
  2. It is only one dataset.
  3. Since the benchmark came out, I made some changes to milk to make it go faster. I hope that other systems do the same, though, so we can have good progress.