ADMEWORKS ModelBuilder - Feature Selection - PSO
This command displays a dialog for Feature Selection by PSO algorithm (Particle Swarm Optimization). The PSO algorithm starts with the creation of a Population of randomly generated Parameter Sets - individuals. This population is called an intelligent swarm. Individuals are then compared according to Objective Function (based on MLR model). The form of objective function favors sets that have the R2 as high as possible, while minimizing the number of Parameters used as Descriptors. The best individual is called Leader. The parameter set is like a location in some kind of space – leader is the individual which is as close to the best location as possible. Leader can change in time because swarm is on the move. PSO finishes when each individual from the swarm is as close to the leader as possible or when the user presses the “Stop” button. After the calculations are stopped, the user may select any of the generated Parameter Sets from the list and make it the Active Set by pressing the “OK” button.