Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master
Commits on Nov 14, 2011
  1. Move CountingProbDist to learning.py and use it in NaiveBayesLearner.…

    authored
    … Make NaiveBayesLearner use target-value frequencies too.
  2. Make BayesNet create its BayesNodes, so they can't possibly be shared…

    authored
    …. (Avoids bug that bit me.)
Commits on Nov 13, 2011
  1. Add repr methods and assertion.

    authored
Commits on Nov 12, 2011
  1. Add doc comments.

    authored
Commits on Nov 11, 2011
Commits on Nov 6, 2011
Commits on Nov 5, 2011
Commits on Nov 4, 2011
  1. Refactor CountingProbDist:

    authored
     * Use weighted_sampler() for the sampling.
     * By not using a defaultdict, keep n_obs always up to date.
     * Don't inherit from ProbDist, at least for now, since we don't
       use anything from the superclass, and if we did it would break
       (e.g. __setitem__).
     * Remove __len__ since nobody uses it.
     * Use heapq.nlargest.
    
    Also separate random doctests from others, and tweak docs.
Commits on Nov 3, 2011
  1. Remove astar row from compare_graph_searchers table because it's sens…

    authored
    …itive to trivial changes. Needs real tests instead.
Commits on Nov 1, 2011
  1. Some more-readable doctests.

    authored
  2. Fill out gibbs_ask().

    authored
Commits on Oct 31, 2011
  1. Partly fill in gibbs_ask().

    authored
  2. Shortened comment.

    authored
Commits on Oct 30, 2011
Commits on Oct 29, 2011
  1. Test all cases for clip().

    authored
  2. Added scalar clip function.

    authored
  3. Fixed search bug reported by Matthew Yurka, and others:

    authored
     * graph_search missed deduping nodes in the frontier
     * astar_search didn't use 3rd-edition pseudocode
     * InstrumentedProblem didn't forward the path_cost method,
       so all the doctest results on instrumented cost-sensitive
       problems were wrong
    
    Also filled in breadth_first_search and per-node memoization 
    of h (to fulfill the doc comment saying that was done).
    
    There's a remaining problem that sometimes trivial refactorings
    change the amount of search that is done, slightly, which should
    not happen since the code is supposed to be deterministic.
Commits on Oct 28, 2011
  1. Don't compare strings with 'is not'. (Broke things in pypy.) Also sho…

    authored
    …w Struct members in sorted order so doctests work in pypy.
Commits on Oct 27, 2011
Commits on Oct 26, 2011
  1. More s/##/#/ for consistency.

    authored
  2. s/successor/child/

    authored
Commits on Oct 25, 2011
  1. Fixed some bugs in AdaBoost.

    authored
Something went wrong with that request. Please try again.