| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | Halfak | T171505 Late-July 2017 ORES deploy | |||
| Resolved | Ladsgroup | T170960 Add new data for damaging models of Persian Wikipedia | |||
| Resolved | Ladsgroup | T171386 Investigate small loss in fitness with the new data in fawiki |
Event Timeline
Comment Actions
Reverted:
ScikitLearnClassifier - type: GradientBoosting - params: init=null, max_leaf_nodes=null, min_samples_split=2, presort="auto", warm_start=false, min_weight_fraction_leaf=0.0, max_features="log2", max_depth=7, learning_rate=0.01, verbose=0, subsample=1.0, loss="deviance", scale=true, n_estimators=500, balanced_sample=false, criterion="friedman_mse", min_samples_leaf=1, center=true, balanced_sample_weight=true, random_state=null, min_impurity_split=1e-07 - version: 0.3.0 - trained: 2017-07-22T07:57:02.413279 Table: ~False ~True ----- -------- ------- False 34858 3228 True 232 934 Accuracy: 0.912 Precision: ----- ----- False 0.993 True 0.224 ----- ----- Recall: ----- ----- False 0.915 True 0.8 ----- ----- PR-AUC: ----- ----- False 0.994 True 0.343 ----- ----- ROC-AUC: ----- ----- False 0.936 True 0.939 ----- ----- Recall @ 0.1 false-positive rate: label threshold recall fpr ------- ----------- -------- ----- False 0.734 0.823 0.097 True 0.445 0.83 0.096 Filter rate @ 0.9 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.572 0.122 0.9 True 0.267 0.802 0.903 Filter rate @ 0.75 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.828 0.27 0.75 True 0.597 0.911 0.754 Recall @ 0.995 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.605 0.88 0.995 True 0.952 0.034 1 Recall @ 0.99 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.33 0.943 0.99 True 0.952 0.034 1 Recall @ 0.98 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.106 0.983 0.98 True 0.952 0.034 1 Recall @ 0.9 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.046 1 0.97 True 0.952 0.036 0.98 Recall @ 0.75 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.046 1 0.97 True 0.948 0.047 0.849 Recall @ 0.6 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.046 1 0.97 True 0.943 0.073 0.672 Recall @ 0.45 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.046 1 0.97 True 0.92 0.225 0.459 Recall @ 0.15 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.046 1 0.97 True 0.301 0.886 0.156
Damaging:
ScikitLearnClassifier - type: GradientBoosting - params: n_estimators=300, warm_start=false, random_state=null, min_samples_leaf=1, balanced_sample=false, criterion="friedman_mse", max_features="log2", init=null, max_leaf_nodes=null, center=true, presort="auto", subsample=1.0, scale=true, min_samples_split=2, verbose=0, balanced_sample_weight=true, max_depth=3, min_weight_fraction_leaf=0.0, min_impurity_split=1e-07, loss="deviance", learning_rate=0.1 - version: 0.3.0 - trained: 2017-07-22T08:13:52.189658 Table: ~False ~True ----- -------- ------- False 34778 3317 True 75 1082 Accuracy: 0.914 Precision: ----- ----- False 0.998 True 0.246 ----- ----- Recall: ----- ----- False 0.913 True 0.935 ----- ----- PR-AUC: ----- ----- False 0.994 True 0.403 ----- ----- ROC-AUC: ----- ----- False 0.964 True 0.973 ----- ----- Recall @ 0.1 false-positive rate: label threshold recall fpr ------- ----------- -------- ----- False 0.396 0.923 0.093 True 0.34 0.969 0.097 Filter rate @ 0.9 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.693 0.125 0.9 True 0.607 0.898 0.905 Filter rate @ 0.75 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.988 0.27 0.752 True 0.818 0.929 0.753 Recall @ 0.995 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.267 0.936 0.995 True 0.987 0.019 1 Recall @ 0.99 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.142 0.958 0.99 True 0.987 0.019 1 Recall @ 0.98 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.054 0.987 0.98 True 0.987 0.019 1 Recall @ 0.9 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.011 1 0.971 True 0.987 0.02 0.983 Recall @ 0.75 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.011 1 0.971 True 0.98 0.075 0.864 Recall @ 0.6 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.011 1 0.971 True 0.975 0.107 0.672 Recall @ 0.45 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.011 1 0.971 True 0.949 0.311 0.479 Recall @ 0.15 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.011 1 0.971 True 0.059 0.996 0.172
Goodfaith:
ScikitLearnClassifier - type: GradientBoosting - params: loss="deviance", init=null, presort="auto", min_weight_fraction_leaf=0.0, min_impurity_split=1e-07, max_depth=7, min_samples_leaf=1, max_features="log2", criterion="friedman_mse", warm_start=false, random_state=null, verbose=0, balanced_sample_weight=true, center=true, n_estimators=500, balanced_sample=false, subsample=1.0, min_samples_split=2, learning_rate=0.01, scale=true, max_leaf_nodes=null - version: 0.3.0 - trained: 2017-07-22T09:52:28.507715 Table: ~False ~True ----- -------- ------- False 582 88 True 3126 35456 Accuracy: 0.918 Precision: ----- ----- False 0.157 True 0.998 ----- ----- Recall: ----- ----- False 0.869 True 0.919 ----- ----- PR-AUC: ----- ----- False 0.242 True 0.995 ----- ----- ROC-AUC: ----- ----- False 0.967 True 0.957 ----- ----- Recall @ 0.1 false-positive rate: label threshold recall fpr ------- ----------- -------- ----- False 0.313 0.943 0.098 True 0.557 0.914 0.093 Filter rate @ 0.9 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.446 0.9 0.907 True 0.712 0.114 0.9 Filter rate @ 0.75 recall: label threshold filter_rate recall ------- ----------- ------------- -------- False 0.68 0.928 0.754 True 0.967 0.263 0.75 Recall @ 0.995 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.954 0.026 1 True 0.3 0.943 0.995 Recall @ 0.99 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.954 0.026 1 True 0.138 0.972 0.99 Recall @ 0.98 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.954 0.026 1 True 0.042 1 0.983 Recall @ 0.9 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.954 0.026 1 True 0.042 1 0.983 Recall @ 0.75 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.95 0.044 0.906 True 0.042 1 0.983 Recall @ 0.6 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.946 0.055 0.786 True 0.042 1 0.983 Recall @ 0.45 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.94 0.085 0.56 True 0.042 1 0.983 Recall @ 0.15 precision: label threshold recall precision ------- ----------- -------- ----------- False 0.423 0.903 0.153 True 0.042 1 0.983