|
|
 |
| |
|
|
|
Evolution of Mathematical Models in Information Systems Project Management: A Systematic Review and Testable Framework |
|
|
|
PP: 1773-1784 |
|
|
doi:10.18576/amis/200620
|
|
|
|
Author(s) |
|
|
|
Umar Ammar Altahtooh,
|
|
|
|
Abstract |
|
|
| This study examines how the mathematical models used to plan, estimate, monitor and optimize information systems projects have evolved, how they are structurally related, and under which project conditions each family of models adds measurable value. A systematic literature review was conducted in accordance with the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement, combining a primary Google Scholar search with Scopus, Web of Science and IEEE Xplore searches, dual independent screening, quality appraisal and citation snowballing; 142 studies were synthesized. The extracted models were grouped into six clusters and integrated, through decision theory, systems theory, contingency theory and operations research, into a formal analytical framework from which three testable propositions were derived. The propositions were tested with reproducible Monte Carlo experiments on 400 randomly generated project networks and 9,000 simulated cost trajectories. Classical network estimates were systematically optimistic: the bias was nil for single-path schedules but reached 8.6% with eight parallel work-streams, and it was driven by the interaction between parallelism and duration uncertainty rather than by uncertainty alone. Committing delivery dates from simulated completion distributions instead of classical network estimates reduced the expected cost of lateness and idle buffers by 19% on average, and by 86% relative to deterministic schedules, but the net benefit became positive only above identifiable thresholds of parallelism, volatility and modelling cost. Earned value cost forecasts lost accuracy in proportion to the volatility of cost efficiency, with mid-project mean absolute errors rising from 0.8% to 7.7%. The results indicate that model choice in information systems projects should be treated as a contingent cost–benefit decision: deterministic models suffice for serial and stable projects, whereas parallel and volatile projects justify simulation-based and adaptive hybrid models. The study converts a descriptive historical classification into an explicit, testable decision framework that researchers can validate empirically and practitioners can use to select quantitative tools. |
|
|
|
|
|
|
|
