| ---
|
| license: cc-by-4.0
|
| task_categories:
|
| - time-series-forecasting
|
| tags:
|
| - tsfile
|
| - timeseries
|
| - time-series
|
| - anomaly-detection
|
| pretty_name: PSM (TsFile)
|
| size_categories:
|
| - 100K<n<1M
|
| configs:
|
| - config_name: default
|
| data_files:
|
| - split: train
|
| path: PSM_train.tsfile
|
| - split: test
|
| path: PSM_test.tsfile
|
| ---
|
|
|
| # PSM (TsFile)
|
|
|
| Apache TsFile version of the `PSM` anomaly-detection subset of
|
| [`thuml/Time-Series-Library`](https://huggingface.co/datasets/thuml/Time-Series-Library).
|
|
|
| ## Overview
|
|
|
| Pooled Server Metrics: 25 server-application metrics for anomaly detection.
|
|
|
| - **Train:** 132,481 rows (all normal).
|
| - **Test:** 87,841 rows (with per-timestep 0/1 anomaly labels).
|
| - **Channels:** 25.
|
|
|
| The train and test segments are stored as two separate TsFiles
|
| (`PSM_train.tsfile` / `PSM_test.tsfile`), preserving the original split.
|
|
|
| ## Schema (TsFile structure)
|
|
|
| - **Time** (INT64, milliseconds) — `timestamp_(min)` * 60000.
|
| - **FIELD** (25 channels, FLOAT) — the sensor/metric channels.
|
| - **label** (INT32) — per-timestep anomaly flag (0/1). The train file is all 0
|
| (no ground-truth labels); the test file carries the anomaly labels.
|
|
|
| No channels or rows are dropped.
|
|
|
| ## Usage
|
|
|
| Read the `.tsfile` files with the Apache TsFile Java or Python SDK.
|
|
|
| ## Source & license
|
|
|
| - Original dataset: https://huggingface.co/datasets/thuml/Time-Series-Library (subset `PSM`)
|
| - Author / publisher: thuml (Tsinghua University)
|
| - Paper: https://arxiv.org/abs/2407.13278
|
| - License: CC BY 4.0
|
|
|
