Skip to main content

Table 4 Results from the online method for both hospitals

From: A unified machine learning approach to time series forecasting applied to demand at emergency departments

algorithm

period

MAE

MAPE (in%)

algorithm

period

MAE

MAPE (in %)

lm

1

14.27

6.7

lm

730

10.59

8.6

glmnet

1

14.31

6.8

lm

1

10.59

8.6

gbm

730

14.33

6.8

glmnet

1

10.60

8.6

lm

7

14.33

6.8

lm

7

10.62

8.6

lm

365

14.34

6.8

gbm

1

10.63

8.6

glmnet

7

14.37

6.8

glmnet

7

10.64

8.6

glmnet

365

14.38

6.8

glmnet

730

10.64

8.5

gbm

7

14.40

6.8

gbm

7

10.71

8.7

lm

60

14.47

6.8

gbm

730

10.78

8.7

glmnet

60

14.47

6.8

lm

365

10.80

8.7

glmnet

1

14.49

6.9

glmnet

365

10.82

8.7

lm

30

14.50

6.8

lm

30

10.84

8.8

glmnet

30

14.50

6.8

glmnet

30

10.84

8.8

gbm

30

14.52

6.9

lm

60

10.86

8.8

gbm

60

14.52

6.9

glmnet

60

10.87

8.8

gbm

365

14.53

6.9

rf

1

10.93

8.9

rf

1

14.55

6.9

gbm

365

10.96

8.9

glmnet

730

14.58

6.8

gbm

30

11.00

8.9

lm

730

14.60

6.9

rf

7

11.08

9.0

rf

7

14.66

6.9

gbm

60

11.15

9.0

rf

730

14.73

6.9

rf

60

11.21

9.1

rf

365

14.76

7.0

rf

30

11.21

9.1

rf

60

15.02

7.1

rf

365

11.49

9.2

rf

30

15.08

7.1

rf

730

11.51

9.0

knn

1

15.52

7.3

knn

7

12.55

10.1

knn

7

15.53

7.3

knn

1

12.60

10.1

knn

730

15.61

7.4

knn

30

12.75

10.2

knn

365

15.62

7.4

knn

60

12.78

10.2

knn

30

15.75

7.4

knn

730

12.81

10.1

knn

60

15.93

7.5

knn

365

12.81

10.1

 

(a) St Mary’s Hospital

   

(b) Charing Cross Hospital