tsdata: Time Series Datasets in Python

This repository is a collection of time series datasets.

The tsdata package itself contains a way of loading these into Pandas.

NOTE: The PyPI name is py-tsdata due to confict with a removed package.

Installing

You can install this as a regular Python package via pip:

pip install py-tsdata

Quickstart

Load a dataset directly into Pandas from a dict-like interface:

>>> from tsdata.fpp3 import raw
>>> "Tourism" in raw
True
>>> raw["Tourism"].head(2)
   Quarter    Region            State   Purpose       Trips
0  1998 Q1  Adelaide  South Australia  Business  135.077690
1  1998 Q2  Adelaide  South Australia  Business  109.987316

Supported Datasets

The currently support datasets are grouped into the following sources:

• tsdata.fpppy, with data from Forecasting: Principles and Practice, the Pythonic Way
• tsdata.fpp3, with data from Forecasting: Principles and Practice, 3rd Edition (extracted from the R package)

Currently only "raw" datasets are implemented, i.e. as-is.

Contributing

If you have time series datasets you would like to add (that you have the rights to contibute), please create a pull request!

Preferred formats are .parquet or .csv, though if it can be read by Pandas - we can add it.