SYNTHETIC POPULATION PRODUCING METHOD FOR PREDICTION SYSTEM OF TRAFFIC NEED

摘要

The present invention relates to a method for generating synthetic population for a traffic demand prediction system. In particular, the present invention relates to a method for generating synthetic population corresponding to reality according to personal complicated properties by relieving disadvantages of a conventional method. To this end, the method of the present invention comprises the following steps: including a synthetic population generating module in the traffic demand prediction system, and setting an initial temperature T by the synthetic population generating module; receiving an inputted cross distribution sample S0 for each feature variable with respect to at least two features; receiving the inputted population number P for each feature variable with respect to at least two features; generating a first sample distribution S1 to satisfy the population number; calculating an absolute error AE1 by AE1 = ¦S1-αS0¦(α>0); calculating a total absolute error TAE1 by TAE1 = ΣAE1; generating a second sample distribution S2 to satisfy the population number; calculating an absolute error AE2 by AE2 = ¦S2-αS0¦(α>0); calculating a total absolute error TAE2 by TAE2 = ΣAE2; calculating an energy difference by ΔE = TAE2 - TAE1; and determining selection of S2 by calculating a selection probability Pr as Pr = 0 when ΔE > 0 and as Pr = exp(ΔE/T) when ΔE < 0.