Kanatoko / XBOS-anomaly-detection

XBOS Anomaly Detection
15 stars 2 forks source link

Error occurs when no feature assign to some clusters #2

Closed SYFT closed 3 years ago

SYFT commented 6 years ago

https://github.com/Kanatoko/XBOS-anomaly-detection/blob/41ee7805cc866c5bcf1850954c790d9b046ae79b/xbos.py#L26

Error occurs when no feature assign to some clusters. kmeans.predict doesn't promise to assign at least one feature to some clusters.

Kanatoko commented 6 years ago

Thank you. Can I ask the size of data points and clusters?

SYFT commented 6 years ago

clusters 25 data point > 8*10^8 This error can be fixed by convert cluster_score to dict. By the way, this implement consump too much memory, I think it could benefit from minibatch-kmeans.

Kanatoko commented 6 years ago

https://github.com/Kanatoko/XBOS-anomaly-detection/commit/f62a253612ed17d8c525a7c78f4a9a34a30793a5

Does this fix the issue?

boukicher commented 6 years ago

There are paper that explain the algorithm in details ?

Kanatoko commented 6 years ago

@boukicher No, there is no paper about XBOS.