Train `reverted` model for cswiki

Closed, ResolvedPublic

Actions

(p3)ladsgroup@ores-compute-01:~/editquality$ make models/cswiki.reverted.rf.model
cut datasets/cswiki.features_reverted.20k_2016.tsv -f2- | \
revscoring train_test \
	revscoring.scorer_models.RF \
	editquality.feature_lists.cswiki.reverted \
	--version 0.0.1 \
	-p 'criterion="entropy"' \
	-p 'max_features="log2"' \
	-p 'n_estimators=640' \
	-p 'min_samples_leaf=3' \
	-s 'table' -s 'accuracy' -s 'precision' -s 'recall' -s 'pr' -s 'roc' -s 'recall_at_fpr(max_fpr=0.10)' -s 'filter_rate_at_recall(min_recall=0.90)' -s 'filter_rate_at_recall(min_recall=0.75)' \
	--balance-sample-weight \
	--center --scale \
	--label-type=bool > \
models/cswiki.reverted.rf.model
2016-06-30 10:04:15,564 INFO:revscoring.utilities.train_test -- Training model...
2016-06-30 10:04:43,739 INFO:revscoring.utilities.train_test -- Testing model...
ScikitLearnClassifier
 - type: RF
 - params: balanced_sample=false, warm_start=false, scale=true, bootstrap=true, n_jobs=1, oob_score=false, min_weight_fraction_leaf=0.0, max_leaf_nodes=null, balanced_sample_weight=true, center=true, verbose=0, max_features="log2", min_samples_split=2, n_estimators=640, min_samples_leaf=3, class_weight=null, max_depth=null, criterion="entropy", random_state=null
 - version: 0.0.1
 - trained: 2016-06-30T10:04:43.736339

Table:
	         ~False    ~True
	-----  --------  -------
	False      3834       30
	True         83       25

Accuracy: 0.972
Precision: 0.455
Recall: 0.231
PR-AUC: 0.322
ROC-AUC: 0.921
Recall @ 0.1 false-positive rate: threshold=None, recall=None, fpr=None
Filter rate @ 0.9 recall: threshold=0.037, filter_rate=0.772, recall=0.907
Filter rate @ 0.75 recall: threshold=0.107, filter_rate=0.89, recall=0.75