Table 1 XGBoost parameters and best parameter combinations (CTRP-L1000 Series Model)
From: Drug-induced cell viability prediction from LINCS-L1000 through WRFEN-XGBoost algorithm
Parameter Name | L1000-CTRP-3h (S1/S2) | L1000-CTRP-6h (S1/S2) | L1000-CTRP-24h (S1/S2) |
|---|---|---|---|
Learning rate | 0.0225/0.0476 | 0.01/0.0225 | 0.01/0.035 |
Gamma | 0/0.0317 | 0.1587/0 | 0/0 |
Max depth | 6/3 | 5/5 | 6/5 |
Min child weight | 4/5 | 3/13 | 8/10 |
Subsample | 0.5757/0.7957 | 0.4343/0.5129 | 0.2457/0.6700 |
Colsample_bytree | 0.1111/0.0794 | 0.4286/0.1270 | 0.4762/0.8095 |
Lambda | 0.01/1.1156 | 1.2103/0.3259 | 1.4946/0.7997 |
Iteration times | 4174/1476 | 5841/3460 | 4968/4492 |