Integrated Feature Selection of ARIMA with Computational Intelligence Approaches for Food Crop Price Prediction

Table 5

The results of ARIMA-MARS for rice price forecasting.

Variable selection results		BF	Coefficient
Variable	Relative importance (%)
Intercept			−13.404
	100.0	BF1 = Max (0, Zt−13 + 35)	0.407
	45.4	BF2 = Max (0, Zt−7 − 0)	23.543
		BF3 = Max (0, Zt−7 + 1.30)	−73.468
		BF4 = Max (0, Zt−7 + 1.87)	49.716
	28.5	BF5 = Max (0, −39.85 − Zt−1)	0.479