MD-MTL 0.0.9

An Ensemble Med-Multi-Task Learning Package

MD-MTL: An Ensemble Med-Multi-Task Learning Package

MD-MTL is a machine learning python package inspired by MALSAR multi-task learning Matlab algorithm, combined with up-to-date multi-task learning researches and algorithm for public research purposes.

Demo

Demo for runing Clustered Multitask Learning algorithm with risk factor analysis, pls copy to your playground and do not ask for change authorizations

Functionality

• Algorithms:
  • Multitask Binary Logistic Regression
    • Hinge Loss
    • L21 normalization
  • Multitask Linear Regression
    • Mean Square Error
    • L21 normalization
  • Cluster Multitask Least Square Regression
    • L21 Normalization
• Util Functions:
  • MTL_data_split
    • Split data set inside each task with predefined proportions, build on sklearn train_test_split
  • MTL_data_extract
    • Extract data from pandas.DataFrame to desired data matrix format, with desired target and task
  • Cross Validation with k Folds:
    • Cross validation with predefined k folds and scoring methods
  • RFA
    • Risk Factor Analysis with Plotly fig returned

more see Documentation

Related Reseaches

Accelerated Gredient Method

Clustered Multi-Task Learning: a Convex Formulation

Regularized Multi-task Learning

Installation (test version)

pip install -i https://test.pypi.org/simple/ MD_MTL==0.0.9

Dependency

Auto generated by pigar

• scikit_learn == 0.22.1

• setuptools == 45.2.0

• tqdm == 4.46.1

• plotly == 4.8.1

• numpy == 1.18.1

• pandas == 1.0.4

• pytest == 5.3.5

• scipy == 1.4.1

Package Update

• Manual Deployment:

  • test-pypi manual

  • python setup.py sdist bdist_wheel

  • twine check dist/*

  • twine upload --repository-url https://test.pypi.org/legacy/ dist/*

  or rewrite .pypirc file with credencials and

  • python3 twine upload -r pypi dist/*

  • python3 setup.py dist bdist_wheel

• Automation(Linux):

  • deploy: ./build_deploy.sh
  • test: ./build_deploy.sh --test

Development

Windows

$ cd Vampyr_MTL

$ python3 -m venv myenv

$ myenv/Scripts/activate

$ pip3 install -r requirements.txt

Doc

https://test.pypi.org/project/MD-MTL/0.0.9/

powered by Sphinx with Google comment style, compile with napoleon:

sphinx-apidoc -f -o docs/source Vampyr_MTL