The Relationship between Cocoa Yield and Climate Variables in Oyo State, Nigeria Using Multiple Linear Regression and Support Vector Machine Analysis

International Journal of Research and Scientific Innovation (IJRSI) | Volume VI, Issue V, May 2019 | ISSN 2321–2705

The Relationship between Cocoa Yield and Climate Variables in Oyo State, Nigeria Using Multiple Linear Regression and Support Vector Machine Analysis

Femi S. Omotayo^1*, Philip G. Oguntunde¹, Olusola S. Makinde ² and Ayorinde. A. Olufayo¹

¹Department of Agricultural and Environmental Engineering, Federal University of Technology, Akure, Nigeria
² Department of Statistics, Federal University of Technology, Akure, Nigeria
*Corresponding author

Abstract: – This study was carried out to evaluate the relationship between cocoa yield and climate variables and examines the variations among them using multiple linear regression, principal component analysis and support vector machine. The climatic and cocoa yield data for thirty (30) years between 1985 and 2014 was used for the study. Cocoa yield exhibits a coefficient of variation of 39.39% with an average of 33637.13 t/ha. The highest coefficient of variation (CV) of 23.4% in the climatic variables was exhibited by wind speed with an annual mean value of 15.6 km/hr which was followed by Pan evaporation with a CV of 12.4% and with an average value of 1396.2 mm/yr. Rainfall has a coefficient of variation (CV) of 16.67% with a mean value of 1355 mm/yr. Solar radiation varied with a coefficient of variation (CV) of 10.3% and an average value of 15.14 MJ/m2/day. The air temperature with a CV of 1.2% has the least variation out of the climatic series that was examined. Cocoa yield temporal trend shows a non-significant decrease at the rate of 0.026 t/ha/yr (P>0.05) decreasing trends. There was a sharp decrease in cocoa yield between 1988 – 1990 and short-term fluctuations of scores of PC1 which runs parallel to those of the yield. A sharp increase in the cocoa yield was noticed between 1990 – 2014. Using SVM regression analysis only rainfall, solar radiation and temperature were the variables that were best correlated with cocoa yield with an r2 value of 0.52. The findings from this research work are expected to provide a baseline for knowledge in regional climate-yield studies. This will aid efforts at assessing high breeds cocoa that could positively respond to future climate actions at mitigating the effect of climate change in the study area. This research work could be extended to other cocoa-producing areas.

Keywords: climate variables; cocoa yield; linear regression; support vector machine, principal component analysis.

I. INTRODUCTION

Cocoa is a major tree crop that has contributed to the growth and development of the economy of Nigeria for a number of years and has gained much relevance because of what people derive from its earnings and its contribution in terms of Gross Domestic Product (GDP) (Oyekale et al., 2009).

Read More