Makefile for ORES Flagged Revisions experiment

Makefile

# See https://phabricator.wikimedia.org/T166235

# From https://quarry.wmflabs.org/query/20200
datasets/fiwiki.flaggedrevs_approved_raw.50k_2017.json:
    wget -qO- https://quarry.wmflabs.org/run/192057/output/0/json-lines?download=true > $@

datasets/fiwiki.flaggedrevs_approved.50k_2017.json: \
        datasets/fiwiki.flaggedrevs_approved_raw.50k_2017.json
    python ~/revscoring/revscoring/utilities/normalize.py < $< > $@

datasets/fiwiki.flaggedrevs_training.65k.json: \
        datasets/fiwiki.labeled_revisions_training.15k_2016.json \
        datasets/fiwiki.flaggedrevs_approved.50k_2017.json
    python ~/revscoring/revscoring/utilities/deduplicate_revs.py $^ > $@

datasets/fiwiki.flaggedrevs_training.w_cache.65k.json: \
        datasets/fiwiki.flaggedrevs_training.65k.json
    cat $< | \
    revscoring extract \
        editquality.feature_lists.fiwiki.reverted \
        editquality.feature_lists.fiwiki.damaging \
        editquality.feature_lists.fiwiki.goodfaith \
        --host https://fi.wikipedia.org \
        --verbose > $@

# FIXME: --observations is working around an old bug in the docopt
models/fiwiki.damaging_w_flaggedrevs_wo_testinfo.gradient_boosting.model: \
        datasets/fiwiki.flaggedrevs_training.w_cache.65k.json
    cat $< | \
    revscoring train_model \
      revscoring.scorer_models.GradientBoosting \
      editquality.feature_lists.fiwiki.damaging \
      damaging \
      --observations "<stdin>" \
      -p 'learning_rate=0.01' \
      -p 'max_features="log2"' \
      -p 'max_depth=5' \
      -p 'n_estimators=700' \
      --balance-sample-weight \
      --version 0.0.1 \
      --center --scale > $@

models/fiwiki.damaging_w_flaggedrevs.gradient_boosting.model: \
        datasets/fiwiki.labeled_revisions_testing.w_cache.5k_2016.json \
        models/fiwiki.damaging_w_flaggedrevs_wo_testinfo.gradient_boosting.model
    revscoring test_model \
    models/fiwiki.damaging_w_flaggedrevs_wo_testinfo.gradient_boosting.model \
    damaging \
    --observations=datasets/fiwiki.labeled_revisions_testing.w_cache.5k_2016.json > $@