adaptiveDesignProcedure

A design procedure of the training data for Machine Learning algorithms able to iteratively add datapoints according to function discrete gradient.

Reference & How to cite:

Most of the theoretical aspects behind adaptiveDesignProcedure are reported in:

M. Bracconi and M. Maestri, "Training set design for Machine Learning techniques applied to the approximation of computationally intensive first-principles kinetic models", Chemical Engineering Journal, 2020, DOI: 10.1016/j.cej.2020.125469

Authors:

adaptiveDesignProcedure is developed and mantained at the Laboratory of Catalysis and Catalytic Processes of Politecnico di Milano by Dr. Mauro Bracconi

Installation:

Clone the repository:

> git clone https://github.com/mbracconi/adaptiveDesignProcedure.git

Change directory:

> cd adaptiveDesignProcedure

To install the package type:

> python setup.py install

To uninstall the package you have to rerun the installation and record the installed files in order to remove them:

> python setup.py install --record installed_files.txt
> cat installed_files.txt | xargs rm -rf

Documentation :

adaptiveDesignProcedure uses Sphinx for code documentation. To build the html versions of the docs simply type:

> cd docs
> make html

Example:

As an example, the "Showcase of the procedure" (Section 4.1 - M. Bracconi & M. Maestri, Chemical Engineering Journal, 2020, DOI: 10.1016/j.cej.2020.125469) is provided in this repository.

Open a terminal and go to example directory:

> cd examples/monodimensional

Run the example:

> python example.py

At the end of the execution, the results of the adaptive procedure are present in the folder.

Requirements:

• numpy
• scipy
• matplotlib
• scikit-learn
• boruta_py

Acknowledgements:

• ERC SHAPE project held by Prof. Matteo Maestri