Document Type

Presentation

Publication Date

7-29-2020

Subjects

Bayesian statistical decision theory, System analysis, Electric power systems -- Models, Regression analysis

Abstract

This talk will describe Reconstructability Analysis (RA), a probabilistic graphical modeling methodology deriving from the 1960s work of Ross Ashby and developed in the systems community in the 1980s and afterwards. RA, based on information theory and graph theory, resembles and partially overlaps Bayesian networks (BN) and log-linear techniques, but also has some unique capabilities. (A paper explaining the relationship between RA and BN will be given in this special session.) RA is designed for exploratory modeling although it can also be used for confirmatory hypothesis testing. In RA modeling, one either predicts some DV from a set of IVs (a directed system), or one discovers relations among a set of variables without making any IV-DV distinction (a neutral system). It can be applied to time-series analysis as well as to spatial patterns. RA can detect high ordinality and nonlinear interactions that are not hypothesized in advance. Unlike neural networks, it is not a black box, but is readily interpretable and explainable. Its graph theoretic conceptual framework allows it to model networks as hypergraphs, and also illuminates in a fundamental way the relationships between wholes and parts, a subject that is central to systems/complexity science. The talk will be an overview of theory and applications of RA, and will introduce OCCAM, a RA software package developed at PSU that is now open source; see https://www.occam-ra.io/ . The web page, http://www.pdx.edu/sysc/research-discrete-multivariate-modeling, documents this methodology, and includes tutorials, published papers, access to the software and some utility programs, and a user’s manual.

Description

Presented at the International Conference on Computational Science (New England Complex Systems Institute) Conference, July 29 2020.

The OCCAM Manual is available online

Persistent Identifier

https://archives.pdx.edu/ds/psu/33725

Share

COinS