Design of experiment

The goal of this repository is to demonstrate the use of DoE. The python notebooks are based on the doepy package and the notebook on logistic regression uses sklearn.

Quadratic function

Firstly we focus on evaluating the coefficient of a quadratic function. Indeed we rarely consider interactions of variables above second or third order. The results can be found in the quadratic_function.ipynb, and noisy_quadratic_function.ipynb that adds a white noise to the evaluation to the quadratic function.

Design of experiments for hyper-parameter search

We then try to apply design of experiment to hyperparameter search in the logistic_regression.ipynb notebook.

Work done for the Missing Data and causality class