mimllearning

miml: Multi-Instance Multi-Label Learning Library for Python

The aim of the library is to ease the development, testing, and comparison of classification algorithms for multi-instance multi-label learning (MIML).

Table of Contents

• Installation
• Documentation
• Usage
• License

Installation

Use the package manager pip to install miml.

$ pip install mimllearning

Requirements

The requirement packages for miml library are: numpy and scikit-learn. Installing miml with the package manager does not install the package dependencies. So install them with the package manager manually if not already downloaded.

$ pip install numpy
$ pip install scikit-learn

Documentation

We can find the documentation of the project in this link: Documentation

Usage

Datasets

from miml.data.load_datasets import load_dataset

dataset_train = load_dataset("miml_birds_random_80train.arff", from_library=True)
dataset_test = load_dataset("C:/Users/Damián/Desktop/miml_birds_random_20test.arff")

Classifier

from miml.classifier import MIMLtoMIBRClassifier, AllPositiveAPRClassifier

classifier_mi = MIMLtoMIBRClassifier(AllPositiveAPRClassifier())
classifier_mi.fit(dataset_train)
results_mi=classifier_mi.evaluate(dataset_test)
probs_mi = classifier_mi.predict_proba(dataset_test)

Report

from miml.report import Report

report = Report(results_mi, probs_mi, dataset_test)
report.to_string()
print("")
report.to_csv()

License

MIML library is released under the GNU General Public License GPLv3.