FTSAD

This is the repository of the paper Future of Time Series Anomaly Detection: From Historical Review to Unified Outlook.

Main Recent Update:

• [Jun. 23, 2025] Project homepage is online.
• [Jul. 03, 2025] (1) Finish the evaluation of metrics. (2) Open supplementary files resource.

Tips

1. All supplementary files can be found at the directory './supp_files'.
2. All datasets can be found in the directories './datasets' or './dataset_tar'. You can use the script './src/data_utils/SimAD_data_loader2.py' to load them. Please note that this script is still under development.
3. To download all the datasets, you should install git-lfs first. Then run git lfs install; git lfs pull to dowanload them.

Acknowledgement

We appreciate the following github repos a lot for their valuable code:

https://github.com/EmorZz1G/SimAD

https://github.com/TheDatumOrg/VUS

https://github.com/ahstat/affiliation-metrics-py

https://github.com/TheDatumOrg/TSB-AD

https://github.com/Raminghorbanii/PATE/

https://github.com/wshw4ng/eTaPR

https://github.com/thuml/Time-Series-Library