Abstract: Factor screening searches for the really important inputs (factors) among the many inputs that are changed in a realistic simulation experiment. Sequential bifurcation (or SB) is a sequential method that changes groups of inputs simultaneously. SB is the most e¢ cient and effective method if the following assumptions are satis ed: (i) second-order polynomials are adequate approximations of the input/output (I/O) functions implied by the simulation model; (ii) the signs of all first-order (or main) effects are known; (iii) if two inputs have no important first-order effects, then they have no important second-order effects either (heredity property). This paper examines SB for random simulation with multiple responses (outputs), called multi-response SB (MSB). This MSB selects "batches" of inputs such that within a batch all inputs have the same sign for a specific type of output, so no cancellation of main effects occurs. MSB also applies Wald's sequential probability ratio test (SPRT) to obtain enough replicates for correctly classifying a group effect or an individual effect as important or unimportant. MSB enables e¢ cient selection of the initial number of replicates in SPRT. The paper also proposes a procedure to validate the three assumptions of MSB. The performance of MSB is examined through extensive Monte Carlo experiments that satisfy all MSB assumptions, and through a case study representing a logistic system in China; MSB performance is very promising.
|Place of Publication||Tilburg|
|Number of pages||37|
|Publication status||Published - 2012|
|Name||CentER Discussion Paper|
- design of experiments
- curse of dimensionality
- sparse effects