Skip to main content

Table 3 The comparison of predictive performance of XGBoost, GBDT, RF, LinearSVM, and DNN on different feature subsets

From: Network analytics and machine learning for predicting length of stay in elderly patients with chronic diseases at point of admission

Models

Metrics

Baseline

Baseline + History

Baseline + MN

Baseline + PSN

Baseline + History + MN + PSN

XGBoost

MAE

4.528 ± 0.006

4.276 ± 0.007

4.300 ± 0.007

4.241 ± 0.007

4.024 ± 0.006

RMSE

6.419 ± 0.013

6.130 ± 0.015

6.182 ± 0.013

6.128 ± 0.013

5.859 ± 0.013

R2

0.250 ± 0.002

0.316 ± 0.002

0.304 ± 0.001

0.316 ± 0.001

0.375 ± 0.002

GBDT

MAE

4.531 ± 0.007

4.280 ± 0.006

4.306 ± 0.006

4.251 ± 0.009

4.026 ± 0.006

RMSE

6.422 ± 0.014

6.136 ± 0.013

6.189 ± 0.012

6.139 ± 0.013

5.861 ± 0.011

R2

0.249 ± 0.002

0.314 ± 0.002

0.302 ± 0.001

0.314 ± 0.001

0.374 ± 0.001

RF

MAE

4.553 ± 0.008

4.343 ± 0.007

4.343 ± 0.006

4.297 ± 0.008

4.106 ± 0.007

RMSE

6.468 ± 0.014

6.229 ± 0.015

6.256 ± 0.013

6.226 ± 0.014

5.987 ± 0.015

R2

0.238 ± 0.002

0.293 ± 0.002

0.287 ± 0.002

0.294 ± 0.001

0.347 ± 0.002

Linear SVM

MAE

4.982 ± 0.007

4.697 ± 0.006

4.714 ± 0.006

4.571 ± 0.007

4.366 ± 0.006

RMSE

7.004 ± 0.011

6.622 ± 0.013

6.710 ± 0.011

6.549 ± 0.012

6.265 ± 0.013

R2

0.107 ± 0.001

0.201 ± 0.002

0.180 ± 0.001

0.219 ± 0.001

0.285 ± 0.001

DNN

MAE

4.595 ± 0.053

4.371 ± 0.043

4.390 ± 0.036

4.302 ± 0.043

4.152 ± 0.046

RMSE

6.518 ± 0.022

6.250 ± 0.020

6.343 ± 0.015

6.223 ± 0.025

6.066 ± 0.034

R2

0.226 ± 0.004

0.289 ± 0.004

0.267 ± 0.003

0.295 ± 0.004

0.330 ± 0.006

  1. The experiment was repeated ten times, and the mean and standard deviation were calculated