Research Article

Novel Two-Stage Method for Low-Order Polynomial Model

Table 4

Mean values of of traditional models (1RS and 2RS) and their corresponding improved models (1IRS and 2IRS) over the 1000 training sets for nine benchmark problems.

ā€‰1RS1IRSDifference2RS2IRSDifference

GP-29.273E+045.052E+04-45.5%4.824E+042.928E+04-39.3%
BH-25.491E+014.358E+01-20.6%3.651E+012.772E+01-24.1%
PM-31.380E+016.063E+00-56.1%3.748E+002.612E+00-30.3%
CP-33.580E+014.259E+00-88.1%4.073E+007.076E-01-82.6%
PS-44.415E+043.098E+04-29.8%1.764E+041.433E+04-18.8%
HM-47.802E-017.274E-01-6.8%4.509E-014.070E-01-9.7%
ZH-109.675E+073.771E+07-61.0%3.375E+079.383E+06-72.2%
DP-153.703E+053.561E+05-3.8%8.920E+048.734E+04-2.1%
WE-201.326E+001.221E+00-7.9%6.240E-016.205E-01-0.6%