Xiao, Chao (2015-08). Comparative Analysis of Entropy Algorithms to Determine the Most Effective Technique for Measuring Complexity in Building Construction. Doctoral Dissertation.
Thesis
Scholars have indicated that construction operation inefficiency is due to particular complexity factors owing to industry specific uncertainties and interdependences. The study of complexity in construction has become an essential topic to provide advanced methods and concepts for construction industry. It also has raised valid questions: Is construction really complex or just complicated? More importantly, how to measure the complexity in building construction systems? This dissertation is based upon these two questions, and intend to fill the research gap that no quantitative complexity measurement has ever been found in research works. Comprehensive literature search is firstly used to make an embedded conceptual analysis of basic concepts of complex and complicated, to conclude building construction systems as complex systems and to metonymic map complex to construction domain. Chaos theory was then used to linked complex building construction systems and entropy complexity measurement together and proposed to use entropy algorithms to measure complexity in building construction. However, entropy in construction could be measured in multiple ways with different results. Therefore, three commonly used entropy algorithms, which are Approximate Entropy, Sample Entropy and Permutation Entropy, were compared along with Six Sigma Analysis and Maximal Lyapunov Exponent based on ten (10) pilots cases and their simulated cases. Two Rounds of simulation were conducted using Monte Carlo Simulation by MATLAB in order to generate more random number to represent different circumstances in building construction performance associate with different sample sizes. The outcomes indicated that the compared with Approximate Entropy and Permutation Entropy, the characteristics of Sample Entropy make be sensitively and efficiently to tell different construction performance circumstances apart by significant complexity measurement for either small sample or large sample. This quantitative measurement of complexity in building construction not only fill the knowledge gap; it also avoids the subjectivity of evaluators and set a unified standard for complexity measurement in building construction in the future research. Understanding complexity in construction management is important for two reasons: (1) to visualize how both complicated and complex traits exist in a construction project (object and social systems), and (2) to identify for stakeholders new types of managerial competencies and tools that reflect the understanding of complexity in construction.
Scholars have indicated that construction operation inefficiency is due to particular complexity factors owing to industry specific uncertainties and interdependences. The study of complexity in construction has become an essential topic to provide advanced methods and concepts for construction industry. It also has raised valid questions: Is construction really complex or just complicated? More importantly, how to measure the complexity in building construction systems?
This dissertation is based upon these two questions, and intend to fill the research gap that no quantitative complexity measurement has ever been found in research works. Comprehensive literature search is firstly used to make an embedded conceptual analysis of basic concepts of complex and complicated, to conclude building construction systems as complex systems and to metonymic map complex to construction domain. Chaos theory was then used to linked complex building construction systems and entropy complexity measurement together and proposed to use entropy algorithms to measure complexity in building construction.
However, entropy in construction could be measured in multiple ways with different results. Therefore, three commonly used entropy algorithms, which are Approximate Entropy, Sample Entropy and Permutation Entropy, were compared along with Six Sigma Analysis and Maximal Lyapunov Exponent based on ten (10) pilots cases and their simulated cases. Two Rounds of simulation were conducted using Monte Carlo Simulation by MATLAB in order to generate more random number to represent different circumstances in building construction performance associate with different sample sizes.
The outcomes indicated that the compared with Approximate Entropy and Permutation Entropy, the characteristics of Sample Entropy make be sensitively and efficiently to tell different construction performance circumstances apart by significant complexity measurement for either small sample or large sample. This quantitative measurement of complexity in building construction not only fill the knowledge gap; it also avoids the subjectivity of evaluators and set a unified standard for complexity measurement in building construction in the future research.
Understanding complexity in construction management is important for two reasons: (1) to visualize how both complicated and complex traits exist in a construction project (object and social systems), and (2) to identify for stakeholders new types of managerial competencies and tools that reflect the understanding of complexity in construction.