rrcf
rrcf copied to clipboard
kLabUM
🌲 Implementation of the Robust Random Cut Forest algorithm for anomaly detection on streams
Hi, and thanks building this great repo! I have a general question; what's the proper way to compute feature importance for RRCF? Basically, I want to know what features contribute...
Hi! I've tested both your implementation of 'streaming detection' and 'batch detection'. So far, I'm getting the best results with the 'batch detection'. However, I want to use the streaming...
I noticed that there are assert statements that are catched wrongly, if an assert statement fails it throws an `AssertionError` not `ValueError` nor `KeyError`. https://github.com/kLabUM/rrcf/blob/34504c14bba233f86a7dcae35d55fc84cc5b7508/rrcf/rrcf.py#L429-L438 Also consider removing all assert...
I ran into issues when a subset of my sample data points only contain ONE unique value. How should we handle such an exception? The error message basically suggests a...
Can RRCF obtain a model from the training set data, and then use this model to detect anomalies in the new data stream?
First, I just want to thanks the maintainer of this project. Second, I want to ask about the current status of this project. Is it still maintained? I have some...
I've tried building a tree from a pandas DataFrame using the following line: `tree = rrcf.RCTree(data_anom.sample(1000, random_state=111))` But it results in a TypeError. It works when I convert it to...
Currently, having any NaN values in the numpy array leads to the following error when trying to build a RCTree: ``` --------------------------------------------------------------------------- ValueError Traceback (most recent call last) in ---->...
Does the implementation support in **timeDecay**? Meaning how much of recent past to consider when computing an anomaly? Same question about **Attribution score** (how much a dimension contributed to anomaly,...
Instead of add, compute, delete, add, compute, delete, ... this PR allows one to directly replace predefined index for LIFO (last in first out) technique. For example, `tree` with 100...