BORF

BORF is now available in the aeon library!

https://www.aeon-toolkit.org/en/stable/api_reference/auto_generated/aeon.transformations.collection.dictionary_based.BORF.html

For a more customizable estimator and for XAI, continue below.

Installation

Install the package using pip by navigating to the directory containing setup.py and running:

pip install .

Basic Usage

Below is an example that demonstrates how to create and use a machine learning pipeline with BORF:

from sklearn.pipeline import make_pipeline
from sklearn.linear_model import RidgeClassifier
import numpy as np
from fast_borf import BorfBuilder
from fast_borf.pipeline.zero_columns_remover import ZeroColumnsRemover
from fast_borf.pipeline.reshaper import ReshapeTo2D
from fast_borf.pipeline.to_scipy import ToScipySparse


# Create a dummy dataset
X = np.random.rand(10, 1, 100)
y = np.random.randint(0, 2, 10)

# Setup the BORF builder
builder = BorfBuilder(
    pipeline_objects=[
        (ZeroColumnsRemover, {}),  # Remove columns with all zeros
        (ReshapeTo2D, {}),  # Reshape the data to 2D
        (ToScipySparse, {}),  # Convert the sparse tensor to a scipy sparse matrix
    ],
)
borf = builder.build(X)
pipe = make_pipeline(borf, RidgeClassifier())

# Transform and train
X_transformed = borf.fit_transform(X)
pipe.fit(X, y)
score = pipe.score(X, y)

Explanation

An example of to get an explanation with BORF can be found in the notebook xai_example.ipynb.