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

Table 10

The results of ARIMA-MARS for wheat price forecasting.

Variable selection results		BF	Coefficient
Variable	Relative importance (%)
Intercept			−13.665
	100.0	BF1 = Max (0, Zt−7 − 26.58)	−1.105
	63.3	BF2 = Max (0, Zt−13 + 26.91)	8.333
		BF3 = Max (0, Zt−13 + 22.55)	−13.064
		BF4 = Max (0, Zt−13 + 17.56)	4.955
	14.4	BF5 = Max (0, Zt−1 + 10.65)	−0.155