Skip to main content

Table 1 Comparison of ARIMA model and hybrid model predictive value PE

From: Medical service demand forecasting using a hybrid model based on ARIMA and self-adaptive filtering method

type of data

CASE

(The amount of data, the amount of observation set)

& ARIMA(p,d,q)

ARIMA forecasting

ARIMA and adaptive combination forecasting

Number of iterations

Iteration time (s)

PEmax

(%)

PEmin(%)

MAPE(%)

σ PE(%)

PEmax

(%)

PEmin

(%)

MAPE(%)

σ PE(%)

Stationarity

Case 1 (348,336)

ARIMA(3, 0, 3)

106.73

0.34

38.4

29.2

0.67

0.00

0.101

0.199

666

2.28

Case 2 (384,372)

ARIMA(2, 0, 5)

134.71

3.25

40.91

35.6

0.62

0.07

0.14

0.19

2208

3

Case 3 (852,840)

ARIMA(7, 0, 7)

49.9

1.8

15.2

13.9

0.192

0.003

0.05

0.06

1670

5

Cyclical and trending

Case 4 (168,156)

ARIMA(5, 1, 5)

11.0

0.1

3.1

2.9

0.08

0.01

0.02

0.02

13,892

7

Case 5 (476,464)

ARIMA(6, 1, 4)

8.25

0.36

4.67

0.23

1.29

0.002

0.42

0.47

30,492

10

Case 6 (252,240)

ARIMA(5, 1, 5)

19.5

1.33

9.1

6.01

2.4

0.0003

0.41

0.74

1410

5

Case 7 (178,168)

ARIMA(5, 1, 5)

3.63

0.33

1.8

1.15

1.266

0.0002

0.509

0.523

14,507

13

Upward trend

Case 8 (82,77)

ARIMA(4, 1, 4)

13.59

2.34

7.33

3.88

0.005

0.004

0.005

0.001

803

3.9

Case 9 (418,406)

ARIMA(3, 1, 6)

3.58

0.17

1.2

0.91

0.103

0.003

0.052

0.029

8932

6.8

Case 10 (43,38)

ARIMA(1, 1, 3)

10.62

6.95

8.5

1.24

8.13

0.07

1.8

3.18

2805

4

Downward trend

Case 11 (304,296)

ARIMA(6, 1, 6)

16.81

2.94

9.78

5.29

1.204

0.002

0.39

0.39

13,050

10

Case 12 (102,92)

ARIMA(1, 0, 0)

4.83

0.36

2.6

1.62

3.62

0.21

1.7

0.99

6622

3.054

Case 13 (65,55)

ARIMA(3, 1, 3)

6.89

1.08

4.46

1.6

5.9

0.64

3.89

1.77

4017

3

  1. ARIMA auto-regressive integrated moving average, PEmax Maximum percentage error, PEmin Minimum percentage error, MAPE mean absolute percentage error, σPE value of the standard deviation of the PE