search

unit8co / darts

A python library for user-friendly forecasting and anomaly detection on time series.
https://unit8co.github.io/darts/
Apache License 2.0
7.54k stars 826 forks source link

show_anomalies for multivariate #2114

Open SaltedfishLZX opened 6 months ago

SaltedfishLZX commented 6 months ago

Hi, I've experimented with the anomaly detection example, and it has proven to be exceptionally helpful in saving me a significant amount of time while establishing the anomaly detection test. I do have a question:

I am currently engaged in multivariate forecasting and wish to conduct anomaly detection on individual components. To achieve this, I have employed component_wise=True for the scorer, and it has been effective. However, when attempting to utilize show_anomalies (which gives visually appealing result figures), all the results are put into a single figure, making it challenging to discern individual results. I have been unable to identify a method to plot each variable separately. Could you provide guidance on this matter?Thanks！

madtoinou commented 6 months ago

Hi @SaltedfishLZX,

This feature is not yet implemented, I added it to the roadmap.

If you have time and would like to contribute to Darts, feel free to tackle it ! :)