Skip to main content

Advertisement

Table 1 Regression RMSE of different methods

From: Random bits regression: a strong general predictor for big data

RMSE Sample Feature Linear KNN NN SVM ELM GBM RF RBR
3D Road Network 434874 2 18.370 6.441 15.548 12.530 16.953 14.819 3.855 2.061
Bike_sharing 17389 16 141.865 104.576 65.994 114.155 94.564 96.765 49.366 40.54
buzz_in_social_media_tomhardware* 28179 97 1.446 0.758 0.373 1.489 1.581 0.311 0.310 0.313
buzz_in_social_media_twitter* 583250 78 1.333 0.516 0.505 - 1.034 0.484 0.471 0.472
computer_hardware 209 7 69.622 63.125 134.912 119.394 159.233 93.214 61.212 50.001
concrete_compressive_strength 1030 9 10.530 8.280 6.355 6.519 13.176 5.823 5.096 3.650
forest_fire* 517 13 1.503 1.399 2.095 1.499 1.401 1.399 1.454 1.390
Housing 506 12 4.884 4.099 4.943 3.752 7.922 3.749 3.097 2.770
istanbul_stock_exchange 536 8 0.012 0.013 0.039 0.013 0.016 0.012 0.013 0.012
parkinsons_telemonitoring 5875 26 9.741 6.097 6.690 7.160 10.354 6.889 3.909 3.954
Physicochemical_properties_of _protein_tertiary_structure 45730 9 5.185 3.790 6.118 6.254 6.118 5.047 3.454 3.407
wine_quality 6497 11 0.736 0.696 0.730 0.676 0.921 0.701 0.585 0.592
yacht_hydrodynamics 308 6 9.134 6.430 1.178 6.542 1.964 1.160 3.833 0.782
year_prediction_MSD 515345 90 9.550 9.216 10.931 - 11.468 9.626 9.242 9.144
  1. The * means the dependent variable of the corresponding data was transformed by log function to be more asymptotically normal
  2. The bold means the first place result of all methods compared