TSDB - Time Series Data Beans

0.8 verified Mon Apr 27 auth: no python

TSDB (Time Series Data Beans) is a Python toolbox that provides easy access to 172 open-source time-series datasets. Current version 0.8, updated September 2023. Release cadence is irregular.

pip install tsdb

Common errors

error ImportError: cannot import name 'load_dataset' from 'tsdb' ↓

cause Outdated version of tsdb (pre-0.8) may not have the load_dataset function.

fix

Upgrade tsdb: pip install --upgrade tsdb

error ValueError: Dataset 'wadi' not found. Available datasets: ... ↓

cause Case-sensitive dataset name; used lowercase instead of correct casing.

fix

Use the exact dataset name as returned by list_datasets().

Warnings

gotcha Dataset names are case-sensitive. Use exact names from the official list (e.g., 'Wadi', not 'wadi'). ↓

fix Check the list of datasets via `from tsdb import list_datasets; print(list_datasets())`.

gotcha Some datasets may require additional download steps (e.g., authentication or manual download). ↓

fix Read the dataset-specific documentation or raise an issue on GitHub.

Imports

load_dataset
wrong
```
from tsdb import TSDB
```
correct
```
from tsdb import load_dataset
```
The main entry point is the load_dataset function, not a class called TSDB.

Quickstart

Load a dataset by name. The function returns a dict with keys 'X_train', 'y_train', etc.

from tsdb import load_dataset
data = load_dataset('Wadi')  # Load a sample dataset
print(data.keys())