Georgina Al Badri: A computational pipeline for analysis of multi parameter models

A talk from the Software Development Processes 2 parallel session at RSECon2022.

=====

Abstract: Computational models involving many parameters, including those based on mathematical models of natural phenomena, require a combination of tools and techniques for effective qualitative and quantitative analysis.

Using a case study of a cell pattern formation model from mathematical biology, this talk will outline a few key considerations and Python tools to facilitate this process. In particular, handling and storing numerous parameter values is considered and addressed using Data Classes, a new feature in Python 3.7. Two open source modules to conduct parameter sensitivity analysis (SALib), and parameter optimisation using the particle swarm method (Pyswarms), are also demonstrated.

Combining tools such as these enable development of an effective computational pipeline to support both model exploration and model parameterisation. This talk will aim to demonstrate how to structure such a pipeline from the ground up, focussing on the early stages of development, to help others handling multi-parameter models.

=====

0:00 Speaker introduction
0:14 Talk
16:16 Q&A