SISSOkit

SISSOkit is a Python library for analysis of SISSO, including generating cross validation files, analyzing results, plotting. Data structures of SISSOkit are mainly numpy array, pandas DataFrame or Series and Python built-in data structure like list, so you can easily build your own code based on SISSOkit.

What is SISSO?

SISSO is short for sure independence screening and sparsifying operator, which is a compressed-sensing method for identifying the best low-dimensional descriptor in an immensity of offered candidates.

References:
R. Ouyang, S. Curtarolo, E. Ahmetcik, M. Scheffler, and L. M. Ghiringhelli, Phys. Rev. Mater. 2, 083802 (2018).
R. Ouyang, E. Ahmetcik, C. Carbogno, M. Scheffler, and L. M. Ghiringhelli, J. Phys.: Mater. 2, 024002 (2019).

For SISSO code, please see SISSO

Getting Started

Dependencies

1. numpy
2. pandas
3. matplotlib

Installation

Using commend line:

pip install SISSOkit

Quick Report

SISSOkit includes some jupyter notebook templates, which you can quickly get a basic analysis of SISSO results without any knowledge about the code. You can find them in the directory notebook_templates.

Or you can simply use function in SISSOkit.notebook. In this case, you only need to specify path to SISSO results, path to which notebook will generate and the notebook name.

For example:

from SISSOkit import notebook

SISSO_path=[
    'path to SISSO results over whole data set',
    'path to cross validation results'
]
notebook_path='notebook path'
notebook_name='regression with CV'

notebook.generate_report(SISSO_path,notebook_path,notebook_name)

Then run all cells, and you will get fundamental analysis of SISSO results.

Usage

Main idea of SISSOkit is that every SISSO result is an instance. The basic class in SISSOkit is Regression, RegressionCV, Classification, ClassificationCV in module SISSOkit.evaluation. To instantiate them, you only need to pass directory path to it.

Arguments in SISSO.in, descriptors, coefficients and intercepts in SISSO.out are all accessible by getting the attributes.

Prediction values, training errors, prediction errors can be acquired by calling the methods.

Documentation

For more detailed information about SISSOkit, please read documentation ./docs/build/html/index.html, which is generated automatically by sphinx.