4 ms·
Dangerously opaque tool. You should tell users why a row is an outlier, and you shoul document the methods you use (even if it’s just links to sklearn). Also wo
by nxpnsv 7y ago
Dangerously opaque tool. You should tell users why a row is an outlier, and you shoul document the methods you use (even if it’s just links to sklearn). Also would be more valuable as a cli app
- je42 7y agohttps://github.com/MateLabs/AutoOut/blob/9c3b1f4195aa96e829cdcde09b688090bac11cc6/app/outlier_treatment/spaces.py https://github.com/MateLabs/AutoOut/blob/9c3b1f4195aa96e829c... Looks like it applies multiple algos: {"model": "zscore", "params": {"threshold":3.5}}, {"model": "DBSCAN", "params": {}}, {"model": "OPTICS", "params": {}}, {"model": "IsolationForest", "params": {}}, {"model": "EllipticEnvelope", "params": {}}, {"model": "OneClassSVM", "params": {}}, {"model": "LocalOutlierFactor", "params": {}},