Welcome to the Multiverse

The archive and benchmark repository for multivariate time series classification.

Datasets · Results · Leaderboard · Evaluation · Classifiers · Contributing

The Multiverse is a new archive of multivariate time series classification datasets. This repository is for accessing, benchmarking, and extending this new archive.

It brings together datasets, published results, reproducible evaluation workflows, and leaderboard infrastructure in one place. The aim is to make it easier to:

• access the multiverse , a collection of benchmark datasets for multivariate time series classification,
• explore and compare against published results of classification algorithms,
• reproduce baseline experiments,
• evaluate new classifiers consistently,
• and contribute new algorithms and results back to the archive.

This repository is intended as both a practical resource for researchers and a public record of benchmark results.

---

Top of the league

Places 1 to 5 by ranks

Further information and more extensive leaderboard views linked here:

• docs/leaderboard.md

Install package

Install the release package from PyPI:

pip install aeon-multiverse

or install the development version from GitHub:

pip install git+https://github.com/aeon-toolkit/multiverse.git

Load a dataset

Use aeon to download data from zenodo and load into memory.

from aeon.datasets import load_classification
X, y = load_classification("BasicMotions")
print(X.shape)
print(y[:10])
trainX, trainy = load_classification("BasicMotions", split="train")
testX, testy = load_classification("BasicMotions", split="test")

More info and links to code - docs/leaderboard.md

Train and test a classifier

Train and test any aeon classifier that can

from aeon.classification.deep_learning import InceptionTimeClassifier
from multiverse.classification import TimesNet

clf = InceptionTimeClassifier()
clf.fit(X, y)
preds = clf.predict(X)

More info and links to aeon classifiers - docs/classifiers.md Multiverse ported classifiers - multiverse/classification

Compare your results to published results

Load results directly in code

from aeon.classification.deep_learning import InceptionTimeClassifier

Or explore published results explored in this repo - [docs/results.md] (docs/results.md)

Run an experiment

To reproduce a benchmark run or evaluate a new classifier, start from:

• experiments/run_single_dataset.py
• experiments/run_benchmark.py

Donate your code and published results

Coming soon

---

Repository layout

multiverse/
├── docs/                  # Documentation
├── experiments/           # Benchmark and reproduction scripts
├── results/               # Submitted results and schema
└── multiverse/            # Python package source for classifiers