Investigation of Plant and Animal Production Values Affecting Consumer Price Index by Multivariate Adaptive Regression Spline: Turkey Case

Abstract

In this study, the outcomes and interpretation of plant and animal production values affecting the consumer price index in Turkey (CPI) were investigated using MARS algorithm. In order to estimate CPI, plant production value (1000 TL), animal production value (1000 TL), livestock value (1000 TL), plant production per capita (TL), animal production per capita (TL), livestock value per capita (TL) variables for 81 provinces in Turkey were used. The compliance criteria were: correlation coefficient r = 0.975, R² = 0.95, Adj. R² = 0.867, GCV = 0.0187, RSS = 1.513, RMSE = 0.137, SDratio = 0.224, MAPE = 1.228, MAD = 0.11, AIC = -222 and AICc = -52. The most significant variables affecting CPI in the increasing direction are basic functions where per capita plant production value (PCPPV) is < 3268 liras, per capita plant production value (PCPPV) is > 1887 liras and per capita livestock value (PCLSV) is > 1766 liras. The most significant variables affecting CPI in the negative direction are basic functions where per capita plant production value (PCPPV) is > 3268 liras, per capita livestock value (PCLSV) is > 1143 liras, and per capita livestock value (PCLSV) is > 1972 liras, respectively. According to these results, it was found that MARS model where interactive variables are also used is an important predictive model for determining the effect of plant and animal production values on other factors.