Back to search

Clemson University

SENSITIVITY AND PRECISION ANALYSIS OF THE GRAPH COMPLEXITY CONNECTIVITY METHOD

Abstract

dc:description.abstract

In the Graph Complexity Connectivity Method (GCCM), twenty nine complexity metrics applied against engineering design graphs are used to create surrogate prediction models of engineering design representations (assembly models and function structures) for given product performance values (assembly time and market value). The performance of these prediction models has been previously assessed solely based on accuracy. In this thesis, the predictive precision of the surrogate models is evaluated in order to assess the GCCM's ability to generate consistent results under the same conditions. The Assembly Model - Assembly Time (AM-AT) prediction model performed the best in terms of both accuracy and precision. This demonstrates that when given assembly models, one can consistently predict accurate assembly times. Further, a sensitivity analysis is conducted to identify the significant complexity metrics in the estimation of the performance values, assembly time and market value. The results of the analysis suggest that for each prediction model, there exists at least one metric from each complexity class (size, interconnection, centrality, and decomposition) which is identified as a significant predictor. Two of the twenty nine complexity metrics are found to be significant for all four prediction models: number of elements and density of the in-core numbers. The significant complexity metrics were used to create simplified surrogate models to predict the product performance values. The test results indicate that the precision of the prediction models increases but the accuracy decreases when the unique significant metric sets are used. Finally, three experiments are conducted in order to investigate the effect of manipulation of the significant complexity metrics in predicting the performance values. The results suggest that the significant metric sets perform better in predicting the product performance values as compared to the manipulated metric sets of either union or intersection of metrics.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MS)
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Mechanical Engineering

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Sridhar, Sudarshan
Contributors dc:contributor
  • Summers, Joshua D.
  • Duarte, Rodrigo M.
  • Kurz, Mary E.

Subjects

dc:subject × 1

Identifiers

dc:identifier.*
Repository record dc:identifier
https://open.clemson.edu/all_theses/2209

Chain of custody

source
Harvested from
Clemson University
Base URL
open.clemson.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Sridhar, Sudarshan. SENSITIVITY AND PRECISION ANALYSIS OF THE GRAPH COMPLEXITY CONNECTIVITY METHOD. Thesis thesis, https://open.clemson.edu/all_theses/2209