Design of Experiments (DOE) with python
An introduction to Design of Experiments (DOE) with python with a simple case study with and without interactions.
Introduction
In this article, I want to explore the topic of Design of Experiments. First of all we should ask ourselves how do we know more about a system? The only way to know more or learn something about a system is to disturb it and then observe it. This is the basis of machine learning actually. Ideally you want your algorithm to learn patterns in your data after all possible disturbances you may have, and with all variables you can find for your experiments. But if you are to design a new experiment to learn from your system you probably can’t afford to wait one year to collect enough data or explore all possible variables combinations in a random manner. There is a scientific way to do it which is proved to give you the best result. This approach or system is called Design of Experiment.
External references:
Most of the content of this article comes from the excellent free course Experimentation for Improvement — Home | Coursera and the website with the free book:
Process Improvement Using Data — Process Improvement using Data (learnche.org)