Conference site » Proceedings

Enabling Active Learning Pedagogy and Insight Mining with a Grammar of Model Analysis

Zachary del Rosario
Assistant Professor of Engineering and Applied Statistics, Olin College of Engineering

Abstract

Modern engineering models are complex, with dozens of inputs, uncertainties arising from simplifying assumptions, and dense output data. While major strides have been made in the computational scalability of complex models, relatively less attention has been paid to user-friendly, reusable tools to explore and make sense of these models. Grama is a python package aimed at supporting these activities. Grama is a grammar of model analysis: an ontology that specifies data (in tidy form), models (with quantified uncertainties), and the verbs that connect these objects. This definition enables a reusable set of evaluation \textquotedbl{}verbs\textquotedbl{} that provide a consistent analysis toolkit across different grama models. This paper presents three case studies that illustrate pedagogy and engineering work with grama: 1. Providing teachable moments through errors for learners, 2. Providing reusable tools to help users self-initiate productive modeling behaviors, and 3. Enabling exploratory model analysis (EMA) – exploratory data analysis augmented with data generation.

Keywords

engineering, engineering education, exploratory model analysis, software design, uncertainty quantification

DOI

10.25080/majora-212e5952-025

Bibtex entry

Full text PDF