machine learning - How to get training accuracy in svmlight with cross validation -
i want run cross validation on training set using svmlight. seems option -x 1 (although i'm not sure how many folds implements...). output is:
xialpha-estimate of error: error<=31.76% (rho=1.00,depth=0) xialpha-estimate of recall: recall=>68.24% (rho=1.00,depth=0) xialpha-estimate of precision: precision=>69.02% (rho=1.00,depth=0) number of kernel evaluations: 56733 computing leave-one-out **lots of gibberish here** retrain on full problem..............done. leave-one-out estimate of error: error=12.46% leave-one-out estimate of recall: recall=86.39% leave-one-out estimate of precision: precision=88.82% actual leave-one-outs computed: 412 (rho=1.00) runtime leave-one-out in cpu-seconds: 0.84
how can accuracy? estimate of error
?
thank you!
these contradicting concepts. training error error on training set, while cross validation used approximate validation error (on data not used training).
your output suggests using n-folds (where n-size of training set) leads called "leave 1 out" validation (only 1 testing point!) overestimating model's quality. should try 10-folds, , accuracy 1-error.
Comments
Post a Comment