---
id: tsdb
version: "0.8"
license: Copyright (c) 2023-present, Wenjie Du All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are… (full text in the JSON record)
license_treatment: permissive
maintenance: active
---

# tsdb — TSDB (Time Series Data Beans): a Python toolbox helping load 172 open-source time-series datasets

License: permissive · Maintenance: active · Downloads: 127.8K/mo

## What it is and what it does

TSDB is a dataset management library that centralizes access to 173 public time-series datasets commonly used in research and benchmarking. It handles downloading, extraction, and caching automatically, letting you load datasets with a single function call. The library is designed to decouple dataset management from machine learning algorithms, so you can experiment with different datasets without writing custom download and preprocessing code.

The package performs minimal data processing—mainly removing invalid samples—and provides both processed and raw dataset access. It manages a local cache (by default in ~/.pypots/tsdb) and lets you list available datasets, delete specific cached items, or migrate the cache directory to external storage. Dependencies are standard data-science libraries (numpy, scipy, pandas, scikit-learn, requests, tqdm), making it straightforward to integrate into existing workflows.

Use it for:

- Benchmark time-series forecasting models against standard datasets like PhysioNet or PeMS traffic data without manual downloading.
- Rapidly prototype imputation and classification algorithms on diverse datasets (air quality, electricity, vessel AIS) with one-line loading.
- Access all 163 UCR & UEA classification datasets for time-series classification research without managing individual downloads.
- Manage dataset caching and storage on resource-constrained machines by migrating the cache directory to external disks.
- Build reproducible research pipelines where dataset versions and availability are centrally maintained by the TSDB project.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

TSDB downloads and loads 173 open-source time-series datasets with minimal setup, handling caching and basic data cleaning automatically so you can focus on analysis.

Yes. TSDB is actively maintained, has no known vulnerabilities, low install friction, and solves a real friction point—downloading and organizing public time-series datasets. It is well-suited for researchers and engineers working on time-series tasks who want to avoid manual dataset management. The permissive BSD license poses no barrier. Install it if you regularly work with benchmark time-series data.

## Install

pip install tsdb
uv add tsdb
poetry add tsdb

## Installing tsdb

Before you install: Low install friction with a pure-Python wheel and seven common dependencies (numpy, scipy, pandas, pyarrow, scikit-learn, requests, tqdm). Active maintenance with recent commits and production-stable status.

License in practice: BSD 3-Clause permissive license allows commercial and private use with minimal restrictions; you must retain copyright notices and disclaimers in source distributions.

Quickstart:

pip install tsdb

import tsdb
data = tsdb.load('physionet_2012')
tsdb.list()  # see all 173 available datasets

Requires Python 3.8 or later; first load of a dataset downloads and caches it locally (default ~/.pypots/tsdb).

Verify before relying:

- Whether all 173 datasets remain consistently available and whether download reliability is documented.
- Performance characteristics when loading large datasets or working with many cached datasets.
- Whether data preprocessing beyond 'removing invalid samples' is actually performed or if raw download is truly the only option.

## Package facts

- License: Copyright (c) 2023-present, Wenjie Du All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are… (full text in the JSON record) (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 127.8K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

time series dataset loader, download time series data, public time series datasets, time series benchmark datasets, tsdb dataset management, forecasting imputation datasets, ucr uea classification data, time-series-data, dataset-management, benchmarking

[View on SkillFed](https://skillfed.io/packages/tsdb) · [View on PyPI](https://pypi.org/project/tsdb/)