TSDB
a Python toolbox loads 173 public time series datasets for machine/deep learning with a single line of code. Datasets from multiple domains including healthcare
What is TSDB?
a Python toolbox loads 173 public time series datasets for machine/deep learning with a single line of code. Datasets from multiple domains including healthcare, financial, power, traffic, weather, and etc.
Overview of TSDB
What it is
TSDB is a Python toolbox that loads 173 public time series datasets for machine learning and deep learning with a single line of code. It is part of the PyPOTS ecosystem and was separated from PyPOTS to decouple datasets from learning algorithms. TSDB covers datasets from multiple domains including healthcare, financial, power, traffic, and weather. The tool helps researchers and engineers avoid data collecting and downloading, allowing them to focus on data processing details.
Key features
- Load 173 public time series datasets with a single line of code
- Datasets from multiple domains: healthcare, financial, power, traffic, weather, and others
- Automatic download, extraction, and processing of datasets
- Caching of loaded datasets with options to manage cache (list, delete, migrate)
- Ability to download and save raw datasets without processing
- Available via pip, conda, or source code installation
Use cases
TSDB is suitable for researchers and engineers working on time series forecasting, imputation, classification, and LLM training. It can be used for benchmarking machine learning and deep learning models on standardized public datasets, or for quickly accessing raw time series data for custom preprocessing pipelines.
Getting started
Install via pip:pip install tsdbOr install from source:pip install https://github.com/WenjieDu/TSDB/archive/main.zipOr install via conda:conda install tsdb -c conda-forgeThen in Python:import tsdb# list all available datasetstsdb.list()# load a datasetdata = tsdb.load('physionet_2012')# download raw datatsdb.download_and_extract('physionet_2012', './save_it_here')# manage cachetsdb.list_cache()tsdb.delete_cache(dataset_name='physionet_2012', only_pickle=True)tsdb.migrate_cache('/mnt/external_disk/TSDB_cache')
License and pricing
License: BSD-3-Clause. Pricing: open source (free).
Based on the project's public README and repository metadata.