Table 2 Parameters settings of GBDT and GRU models.

From: Estimated date of delivery with electronic medical records by a hybrid GBDT-GRU model

Model

Parameters

Values

GBDT

Learning rate

0.01

Loss

“Ls”

N_estimators

500

Min_samples_leaf

4

Min_samples_split

3

Max_depth

3

GRU

Batch size

100

Loss function

MSE

Layers

1

Optimizer

Adam

Hidden_size

37

Input size

18

Learning rate

0.002

Epochs

200