A Comparative Analysis of Traditional SARIMA and Machine Learning Models for CPI Data Modelling in Pakistan

Table 3

Estimated candidate models for CPI (%) from the year 1960 to 2021.

Candidate models	RMSE	MSE	MAPE
SARIMA (2, 1, 2) (0, 1, 0)12	2.06	4.24	2.201
SARIMA (2, 1, 2) (1, 1, 0)12	1.91	3.64	1.890
SARIMA (3, 1, 3) (0, 1, 0)12	1.88	3.55	2.188
NNAR (iterations = 20)	2.56	6.55	6.970
NNAR (iterations = 30)	2.59	6.71	7.054
NNAR (iterations = 40)	2.55	6.50	7.142
MLP with 5 hidden layers	1.88	3.54	0.031
MLP with 10 hidden layers	1.76	3.09	0.024
MLP with 20 hidden layers	1.32	1.75	0.021