Feature selection using XGBoost
Examples
library(dplyr)
meta_dat1 <- t(meta_dat) %>%
as.data.frame() %>%
dplyr::mutate(group=group)
result_ML_xgboost <- ML_xgboost(meta_dat1)
#> [1] train-rmse:0.364289 test-rmse:0.472248
#> [2] train-rmse:0.265516 test-rmse:0.429406
#> [3] train-rmse:0.193648 test-rmse:0.398691
#> [4] train-rmse:0.141388 test-rmse:0.385973
#> [5] train-rmse:0.103498 test-rmse:0.380957
#> [6] train-rmse:0.076135 test-rmse:0.373863
#> [7] train-rmse:0.056617 test-rmse:0.370007
#> [8] train-rmse:0.042568 test-rmse:0.368762
#> [9] train-rmse:0.031942 test-rmse:0.374442
#> [10] train-rmse:0.024360 test-rmse:0.372204
#> [11] train-rmse:0.018757 test-rmse:0.371477
#> [12] train-rmse:0.014496 test-rmse:0.374902
#> [13] train-rmse:0.011381 test-rmse:0.374174
#> [14] train-rmse:0.008995 test-rmse:0.374225
#> [15] train-rmse:0.007213 test-rmse:0.373786
#> [16] train-rmse:0.005855 test-rmse:0.373830
#> [17] train-rmse:0.004803 test-rmse:0.374031
#> [18] train-rmse:0.003961 test-rmse:0.373514
#> [19] train-rmse:0.003288 test-rmse:0.373618
#> [20] train-rmse:0.002727 test-rmse:0.374203
#> [21] train-rmse:0.002277 test-rmse:0.374671
#> [22] train-rmse:0.001915 test-rmse:0.375046
#> [23] train-rmse:0.001621 test-rmse:0.375243
#> [24] train-rmse:0.001381 test-rmse:0.375285
#> [25] train-rmse:0.001183 test-rmse:0.375266
#> [26] train-rmse:0.001023 test-rmse:0.375258
#> [27] train-rmse:0.000881 test-rmse:0.375279
#> [28] train-rmse:0.000765 test-rmse:0.375267
#> [29] train-rmse:0.000662 test-rmse:0.375142
#> [30] train-rmse:0.000572 test-rmse:0.375137
#> [31] train-rmse:0.000515 test-rmse:0.375145
#> [32] train-rmse:0.000471 test-rmse:0.375150
#> [33] train-rmse:0.000436 test-rmse:0.375157
#> [34] train-rmse:0.000407 test-rmse:0.375101
#> [35] train-rmse:0.000378 test-rmse:0.375086
#> [36] train-rmse:0.000378 test-rmse:0.375084
#> [37] train-rmse:0.000378 test-rmse:0.375082
#> [38] train-rmse:0.000378 test-rmse:0.375080
#> [39] train-rmse:0.000378 test-rmse:0.375079
#> [40] train-rmse:0.000378 test-rmse:0.375078
#> [41] train-rmse:0.000378 test-rmse:0.375078
#> [42] train-rmse:0.000378 test-rmse:0.375077
#> [43] train-rmse:0.000378 test-rmse:0.375077
#> [44] train-rmse:0.000378 test-rmse:0.375077
#> [45] train-rmse:0.000378 test-rmse:0.375077
#> [46] train-rmse:0.000378 test-rmse:0.375077
#> [47] train-rmse:0.000378 test-rmse:0.375077
#> [48] train-rmse:0.000378 test-rmse:0.375077
#> [49] train-rmse:0.000378 test-rmse:0.375077
#> [50] train-rmse:0.000378 test-rmse:0.375077
#> [51] train-rmse:0.000378 test-rmse:0.375077
#> [52] train-rmse:0.000378 test-rmse:0.375077
#> [53] train-rmse:0.000378 test-rmse:0.375077
#> [54] train-rmse:0.000378 test-rmse:0.375077
#> [55] train-rmse:0.000378 test-rmse:0.375077
#> [56] train-rmse:0.000378 test-rmse:0.375077
#> [57] train-rmse:0.000378 test-rmse:0.375077
#> [58] train-rmse:0.000378 test-rmse:0.375077
#> [59] train-rmse:0.000378 test-rmse:0.375077
#> [60] train-rmse:0.000378 test-rmse:0.375077
#> [61] train-rmse:0.000378 test-rmse:0.375077
#> [62] train-rmse:0.000378 test-rmse:0.375077
#> [63] train-rmse:0.000378 test-rmse:0.375077
#> [64] train-rmse:0.000378 test-rmse:0.375077
#> [65] train-rmse:0.000378 test-rmse:0.375077
#> [66] train-rmse:0.000378 test-rmse:0.375077
#> [67] train-rmse:0.000378 test-rmse:0.375077
#> [68] train-rmse:0.000378 test-rmse:0.375077
#> [69] train-rmse:0.000378 test-rmse:0.375077
#> [70] train-rmse:0.000378 test-rmse:0.375077
result_ML_xgboost$p
result_ML_xgboost$feature_result
#> Feature Gain Cover Frequency Importance
#> <fctr> <num> <num> <num> <num>
#> 1: C02630 4.115534e-01 0.07836991 0.05882353 4.115534e-01
#> 2: C18170 2.185850e-01 0.07836991 0.05882353 2.185850e-01
#> 3: C00365 1.161917e-01 0.07836991 0.05882353 1.161917e-01
#> 4: C05581 7.086497e-02 0.12225705 0.17647059 7.086497e-02
#> 5: C02291 6.647211e-02 0.15673981 0.11764706 6.647211e-02
#> 6: C00408 4.876780e-02 0.08150470 0.11764706 4.876780e-02
#> 7: C05674 3.237925e-02 0.07836991 0.05882353 3.237925e-02
#> 8: C02985 2.609616e-02 0.15673981 0.11764706 2.609616e-02
#> 9: C03264 4.927159e-03 0.04388715 0.05882353 4.927159e-03
#> 10: C01250 4.125904e-03 0.08777429 0.11764706 4.125904e-03
#> 11: C05635 3.651393e-05 0.03761755 0.05882353 3.651393e-05