The Genetic Algorithm provides a sophisticated method to allocate a limited set of resources across hundreds or thousands of alternatives in the most value- intensive manner. It accounts for multiple budget pools, interdependencies of projects, and multiple time-periods among other factors, and it provides scenarios for the best resource allocation outcomes for the organization.
The optimization solution in Decision Lens has to account for the following challenges:
• Multiple alternatives requesting resources,
• Over multiple time periods, • Where each alternative has the capability to “float” or move between time periods (the “sequencing”),
• Constrained by limiting factors such as: > Forced selection in specific time periods, > Logical dependencies, > Multiple types and pools of resources, > Matching rules between resource pools, and > Requirements to fully resource alternatives
For simple cases involving only one time period or not very many alternatives, a simple linear programming solution could be sufficient to find the optimal resource allocation. Linear programming on a larger scale is too slow to be useful - a model involving 100 alternatives with funding over 5 years that allows alternatives to “float” between time periods can take hours to solve. Models with 1000 alternatives take days, or even weeks.
The Genetic Algorithm (GA) solves this by mimicking evolutionary processes to create and evolve populations of solutions to find the best answer in a very short amount of time. Our GA has Chromosomes, Evolution, Fitness, Initial Population, and finally Termination that represents the best output. Benefit-Cost Ratio: The simplest assessment of value that goes into the optimizer’s calculations is the Cost-Benefit Ratio, which is the alternative’s priority score divided into its cost. A higher benefit-cost ratio in general means that a project should have better “bang for the buck” and be more likely to be funded under an optimization scenario.