Student Enrollment and Teacher Statistics Forecasting Based on Time-Series Analysis
Table 3
Performance comparison of different forecasting models for student enrollment dataset.
Number of students
Criteria
ARIMA
ETS
TBATS
GRIDSVR
PSOSVR
WOASVR
Primary school
Public
MAPE (%)
3.57
12.89
3.11
6.65
1.61
0.86
RMSE
47989.51
159243.65
40320.54
86053.33
21606.28
12787.62
Private
MAPE (%)
3.12
7.16
7.95
6.06
1.46
1.07
RMSE
1243.26
2626.75
3439.16
2761.55
739.57
528.48
Secondary school
Public
MAPE (%)
6.03
15.59
16.53
8.74
6.00
5.57
RMSE
40912.00
98097.44
116066.74
74626.56
40615.21
39029.75
Private
MAPE (%)
2.14
3.40
3.72
3.61
2.46
1.85
RMSE
2233.16
3421.01
4051.42
3522.10
2418.49
2049.95
High school
Public
MAPE (%)
2.15
2.55
2.03
4.80
1.95
1.71
RMSE
12232.91
14061.67
11797.81
22763.48
12199.88
11762.71
Private
MAPE (%)
7.68
7.51
7.55
7.56
6.75
4.16
RMSE
33472.75
30483.37
32005.37
32036.26
30279.43
17782.32
University
Public
MAPE (%)
0.24
0.89
0.91
1.61
0.22
0.2
RMSE
1181.90
4188.08
4933.25
7319.07
1208.03
1105.6
Private
MAPE (%)
5.06
3.56
9.43
4.87
3.80
1.36
RMSE
50997.54
39093.13
95690.97
56050.40
48495.19
15621.16
Average
Public
MAPE (%)
3.00
7.98
5.65
5.45
2.45
2.09
RMSE
25579.08
68897.71
43279.59
47690.61
18907.35
16171.42
Private
MAPE (%)
4.50
5.41
7.16
5.53
3.62
2.11
RMSE
21986.68
18906.07
33796.73
23592.58
20483.17
8995.48
ARIMA, autoregressive integrated moving average; ETS, exponential smoothing; TBATS, Trigonometric Seasonal Box–Cox Transformation with ARMA residuals Trend and Seasonal Components; GRIDSVR, grid search support vector regression; PSOSVR, particle swarm optimization support vector regression; WOASVR, whale optimization algorithm support vector regression; boldface, the best values in each row.