On 12/13/2014 01:09 AM, He-chien Tsai wrote: > Thanks for reply. I misused random.seed as it returns None. > I passed an integer to random_state but it remains that unexpected > behaviors. > After I cloned the estimator by sklearn.base.clone,e the result > becomes reasonable. > > clfs = [ (clone(pipe).fit(x[train_index], y[train_index]), > (x[test_index], y[test_index])) for > train_index, test_index in KFold(x.shape[0], n_folds=8, > shuffle=True, random_state=254)] > scores = [m.accuracy_score(p[1][1], p[0].predict(p[1][0])) for p in clfs] > > It looks like the estimator keep previous states when they're training That should definitely not happen and would be a serious bug. If you can provide a minimal reproduceable example, please open an issue on github.
------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=164703151&iu=/4140/ostg.clktrk _______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
