Table 2 Evaluating optimal gradient boosting hyperparameters using four optimization techniques within a specified parameter range.

From: Developing a cost-effective tool for choke flow rate prediction in sub-critical oil wells using wellhead data

Tuning parameter

Considered range

SADE

ES

BPI

BBO

n_estimators

[50–300]

223

159

247

230

max_depth

[5–20]

5

15

14

12

max_features

[0.1–1]

0.7146

0.2846

0.2221

0.2595

min_samples_split

[0.01–0.5]

0.2850

0.3593

0.2522

0.3079

learning_rate

[0.01–0.3]

0.0191

0.0335

0.0339

0.0359

subsample

[0.5–1]

0.6535

0.9836

0.6462

0.7706

min_samples_leaf

[0.01–0.5]

3.0000

0.0173

0.0206

0.0191

Back to article page