Tutorial 0: Getting Started

This tutorial takes you through a basic working example of how to use this codebase, including all the different components, up to the results generation. If you’d like to know about the statistics and plotting, see the next tutorial.

# Authors: Vinay Jayaram <vinayjayaram13@gmail.com>
#
# License: BSD (3-clause)

Introduction

To use the codebase you need an evaluation and a paradigm, some algorithms, and a list of datasets to run it all on. You can find those in the following submodules; detailed tutorials are given for each of them.

import numpy as np
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis as LDA
from sklearn.metrics import accuracy_score, roc_auc_score
from sklearn.model_selection import GridSearchCV
from sklearn.pipeline import make_pipeline
from sklearn.svm import SVC

If you would like to specify the logging level when it is running, you can use the standard python logging commands through the top-level moabb module

import moabb
from moabb.datasets import BNCI2014_001, utils
from moabb.evaluations import CrossSessionEvaluation
from moabb.paradigms import LeftRightImagery
from moabb.pipelines.features import LogVariance

In order to create pipelines within a script, you will likely need at least the make_pipeline function. They can also be specified via a .yml file. Here we will make a couple pipelines just for convenience

moabb.set_log_level("info")

Create pipelines

We create two pipelines: channel-wise log variance followed by LDA, and channel-wise log variance followed by a cross-validated SVM (note that a cross-validation via scikit-learn cannot be described in a .yml file). For later in the process, the pipelines need to be in a dictionary where the key is the name of the pipeline and the value is the Pipeline object

pipelines = {}
pipelines["AM+LDA"] = make_pipeline(LogVariance(), LDA())
parameters = {"C": np.logspace(-2, 2, 10)}
clf = GridSearchCV(SVC(kernel="linear"), parameters)
pipe = make_pipeline(LogVariance(), clf)

pipelines["AM+SVM"] = pipe

Datasets

Datasets can be specified in many ways: Each paradigm has a property ‘datasets’ which returns the datasets that are appropriate for that paradigm

print(LeftRightImagery().datasets)

[<moabb.datasets.bnci.bnci_2014.BNCI2014_001 object at 0x7f42e8153220>, <moabb.datasets.bnci.bnci_2014.BNCI2014_004 object at 0x7f42daf31f60>, <moabb.datasets.beetl.Beetl2021_A object at 0x7f42db47b550>, <moabb.datasets.beetl.Beetl2021_B object at 0x7f42daf31ea0>, <moabb.datasets.brandl2020.Brandl2020 object at 0x7f42daf32200>, <moabb.datasets.chang2025.Chang2025 object at 0x7f42daf32260>, <moabb.datasets.gigadb.Cho2017 object at 0x7f42daf321d0>, <moabb.datasets.dreyer2023.Dreyer2023 object at 0x7f42daf32170>, <moabb.datasets.dreyer2023.Dreyer2023A object at 0x7f42daf32080>, <moabb.datasets.dreyer2023.Dreyer2023B object at 0x7f42daf31c90>, <moabb.datasets.dreyer2023.Dreyer2023C object at 0x7f42daf314b0>, <moabb.datasets.forenzo2023.Forenzo2023 object at 0x7f42daf31c60>, <moabb.datasets.mpi_mi.GrosseWentrup2009 object at 0x7f42daf31cc0>, <moabb.datasets.guttmann_flury2025.GuttmannFlury2025_ME object at 0x7f42daf31e40>, <moabb.datasets.guttmann_flury2025.GuttmannFlury2025_MI object at 0x7f42daf315d0>, <moabb.datasets.hefmi_ich2025.HefmiIch2025 object at 0x7f42daf315a0>, <moabb.datasets.kaya2018.Kaya2018 object at 0x7f42daf31b40>, <moabb.datasets.kumar2024.Kumar2024 object at 0x7f42daf31f90>, <moabb.datasets.Lee2019.Lee2019_MI object at 0x7f42daf320e0>, <moabb.datasets.liu2024.Liu2024 object at 0x7f42daf31db0>, <moabb.datasets.physionet_mi.PhysionetMI object at 0x7f42daf31bd0>, <moabb.datasets.schirrmeister2017.Schirrmeister2017 object at 0x7f42daf31d80>, <moabb.datasets.bbci_eeg_fnirs.Shin2017A object at 0x7f42daf31990>, <moabb.datasets.stieger2021.Stieger2021 object at 0x7f42daf319c0>, <moabb.datasets.wairagkar2018.Wairagkar2018 object at 0x7f42daf31a80>, <moabb.datasets.Weibo2014.Weibo2014 object at 0x7f42daf31570>, <moabb.datasets.wu2020.Wu2020 object at 0x7f42daf31750>, <moabb.datasets.yang2025.Yang2025 object at 0x7f42daf317e0>, <moabb.datasets.Zhou2016.Zhou2016 object at 0x7f42daf31ae0>, <moabb.datasets.zhou2020.Zhou2020 object at 0x7f42daf31960>]

Or you can run a search through the available datasets:

print(utils.dataset_search(paradigm="imagery", min_subjects=6))

[<moabb.datasets.alex_mi.AlexMI object at 0x7f42daf31f90>, <moabb.datasets.bnci.bnci_2014.BNCI2014_001 object at 0x7f42daf315d0>, <moabb.datasets.bnci.bnci_2014.BNCI2014_002 object at 0x7f42daf315a0>, <moabb.datasets.bnci.bnci_2014.BNCI2014_004 object at 0x7f42daf31de0>, <moabb.datasets.bnci.bnci_2015.BNCI2015_001 object at 0x7f42daf31b40>, <moabb.datasets.bnci.bnci_2015.BNCI2015_004 object at 0x7f42daf31e40>, <moabb.datasets.bnci.bnci_2019.BNCI2019_001 object at 0x7f42daf31cc0>, <moabb.datasets.bnci.bnci_2020.BNCI2020_001 object at 0x7f42daf32110>, <moabb.datasets.bnci.bnci_2022_001.BNCI2022_001 object at 0x7f42daf31c60>, <moabb.datasets.bnci.bnci_2024_001.BNCI2024_001 object at 0x7f42daf321d0>, <moabb.datasets.bnci.bnci_2025.BNCI2025_001 object at 0x7f42daf32170>, <moabb.datasets.bnci.bnci_2025.BNCI2025_002 object at 0x7f42daf314b0>, <moabb.datasets.brandl2020.Brandl2020 object at 0x7f42daf32080>, <moabb.datasets.chang2025.Chang2025 object at 0x7f42daf31f60>, <moabb.datasets.gigadb.Cho2017 object at 0x7f42daf31c90>, <moabb.datasets.dreyer2023.Dreyer2023 object at 0x7f42daf31bd0>, <moabb.datasets.dreyer2023.Dreyer2023A object at 0x7f42daf31ea0>, <moabb.datasets.dreyer2023.Dreyer2023B object at 0x7f42daf31d80>, <moabb.datasets.dreyer2023.Dreyer2023C object at 0x7f42daf319c0>, <moabb.datasets.fake.FakeDataset object at 0x7f42daf31570>, <moabb.datasets.forenzo2023.Forenzo2023 object at 0x7f42daf31a80>, <moabb.datasets.gao2026.Gao2026 object at 0x7f42daf31990>, <moabb.datasets.mpi_mi.GrosseWentrup2009 object at 0x7f42daf31750>, <moabb.datasets.guttmann_flury2025.GuttmannFlury2025_ME object at 0x7f42daf317e0>, <moabb.datasets.guttmann_flury2025.GuttmannFlury2025_MI object at 0x7f42daf31960>, <moabb.datasets.hefmi_ich2025.HefmiIch2025 object at 0x7f42daf320b0>, <moabb.datasets.jeong2020.Jeong2020 object at 0x7f42daf31ae0>, <moabb.datasets.kaya2018.Kaya2018 object at 0x7f42daf31c30>, <moabb.datasets.kumar2024.Kumar2024 object at 0x7f42daf32020>, <moabb.datasets.Lee2019.Lee2019_MI object at 0x7f42daf31ba0>, <moabb.datasets.liu2024.Liu2024 object at 0x7f42daf32230>, <moabb.datasets.liu2025.Liu2025 object at 0x7f42daf318d0>, <moabb.datasets.ma2020.Ma2020 object at 0x7f42daf318a0>, <moabb.datasets.upper_limb.Ofner2017 object at 0x7f42daf31600>, <moabb.datasets.physionet_mi.PhysionetMI object at 0x7f42daf32050>, <moabb.datasets.rozado2015.Rozado2015 object at 0x7f42daf32140>, <moabb.datasets.schirrmeister2017.Schirrmeister2017 object at 0x7f42daf31ab0>, <moabb.datasets.bbci_eeg_fnirs.Shin2017A object at 0x7f42daf31fc0>, <moabb.datasets.bbci_eeg_fnirs.Shin2017B object at 0x7f42daf31cf0>, <moabb.datasets.stieger2021.Stieger2021 object at 0x7f42daf31ff0>, <moabb.datasets.tavakolan2017.Tavakolan2017 object at 0x7f42daf319f0>, <moabb.datasets.triana_guzman2024.TrianaGuzman2024 object at 0x7f42daf321a0>, <moabb.datasets.wairagkar2018.Wairagkar2018 object at 0x7f42daf31a50>, <moabb.datasets.Weibo2014.Weibo2014 object at 0x7f42daf31b70>, <moabb.datasets.wu2020.Wu2020 object at 0x7f42daf31660>, <moabb.datasets.yang2025.Yang2025 object at 0x7f42daf32290>, <moabb.datasets.yi2025.Yi2025 object at 0x7f42daf309a0>, <moabb.datasets.zhang2017.Zhang2017 object at 0x7f42daf322f0>, <moabb.datasets.zhou2020.Zhou2020 object at 0x7f42daf31840>, <moabb.datasets.zuo2025.Zuo2025 object at 0x7f42e8152a70>]

Or you can simply make your own list (which we do here due to computational constraints)

dataset = BNCI2014_001()
dataset.subject_list = dataset.subject_list[:2]
datasets = [dataset]

Paradigm

Paradigms define the events, epoch time, bandpass, and other preprocessing parameters. They have defaults that you can read in the documentation, or you can simply set them as we do here. A single paradigm defines a method for going from continuous data to trial data of a fixed size. To learn more look at the tutorial Exploring Paradigms

fmin = 8
fmax = 35
# You can inject custom scoring directly into the paradigm (single or multi-metric).
custom_scorer = [
    accuracy_score,
    (roc_auc_score, {"needs_threshold": True}),
]
paradigm = LeftRightImagery(fmin=fmin, fmax=fmax, scorer=custom_scorer)

Evaluation

An evaluation defines how the training and test sets are chosen. This could be cross-validated within a single recording, or across days, or sessions, or subjects. This also is the correct place to specify multiple threads.

evaluation = CrossSessionEvaluation(
    paradigm=paradigm, datasets=datasets, suffix="examples", overwrite=False
)
results = evaluation.process(pipelines)

[codecarbon WARNING @ 19:50:44] Multiple instances of codecarbon are allowed to run at the same time.
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+LDA | BNCI2014-001 | 1 | 0train: Score 0.729
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+SVM | BNCI2014-001 | 1 | 0train: Score 0.743
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+LDA | BNCI2014-001 | 1 | 1test: Score 0.715
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+SVM | BNCI2014-001 | 1 | 1test: Score 0.715
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+LDA | BNCI2014-001 | 2 | 0train: Score 0.597
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+SVM | BNCI2014-001 | 2 | 0train: Score 0.500
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+LDA | BNCI2014-001 | 2 | 1test: Score 0.521
2026-03-21 19:51:11,094 INFO MainThread moabb.evaluations.base AM+SVM | BNCI2014-001 | 2 | 1test: Score 0.500
/home/runner/work/moabb/moabb/moabb/analysis/results.py:192: H5pyDeprecationWarning: Creating a dataset without passing data or dtype is deprecated. Pass an explicit dtype. Using dtype='f4' will keep the current default behaviour.
  dset.create_dataset(

Results are returned as a pandas DataFrame. When multiple metrics are provided, MOABB adds a primary score plus one column per metric (e.g., score_accuracy_score, score_roc_auc_score).

print(results.head())

      score      time  ...  pipeline                  codecarbon_task_name
0  0.729167  0.012792  ...    AM+LDA  47baa456-b73b-40b4-8266-02c32129b6d1
1  0.715278  0.010067  ...    AM+LDA  3a0c5c1d-49c5-4813-b5ed-dc7f932dd037
2  0.597222  0.009965  ...    AM+LDA  1096b1c0-f313-40b9-ae10-37e8f013e101
3  0.520833  0.010062  ...    AM+LDA  9683154a-57b7-4acb-9eac-bb4d257c875d
4  0.743056  0.135124  ...    AM+SVM  dd0b960c-7ca7-41ef-87cf-6dd87d40c54d

[5 rows x 15 columns]