Authors :
David Umolo; George. O. Edeki
Volume/Issue :
Volume 8 - 2023, Issue 11 - November
Google Scholar :
http://tinyurl.com/45z75jk5
Scribd :
http://tinyurl.com/3fw3p662
DOI :
https://doi.org/10.5281/zenodo.10469382
Abstract :
This work investigates the impact of
temperature and rainfall on the yield of cocoa using an
experimental dataset from the Cocoa Research Institute
of Nigeria, Ibadan (n=213) with a blocking factor having
4 levels.
The data were analyzed using exploratory data
analysis and the response surface methodology. The
exploratory data analysis relationship/distributional plot
shows that there exists significant negative relationship
between the yield of cocoa and the predictors
(temperature and rainfall). The estimated boxplot with
respect to blocking factor indicates that there is presence
of outlier in the yield of cocoa with majority of the yields
measured below 500kg over the period of study.
Results from the response surface models without
blocking indicate that all the estimated models were
statistically significant with all the lack of fit test
estimated to be insignificant (an indication of good fit).
On the basis of incorporated blocking factor to the
experiment, we observed that all models which range
from first order to second order outperformed those
without blocking factor by considering the estimated
adjusted R
2
. The blocking factors incorporated into the
experiment were found to be statistically significant with
all contour plots on the basis of the Eigen analysis
suggesting insignificant lack of fit. This implies that
incorporating blocking factor helped minimize the sum
of squared error and in turn improved the precision.This
study recommends that CRIN and other cocoa farmers
should learn to adopt newly developed techniques that
could militate against the impact of weather change
being experienced.
Keywords :
Response surface model, cocoa yield, rainfall, temperature, Eigen-analysis, contour, lack of fit.
This work investigates the impact of
temperature and rainfall on the yield of cocoa using an
experimental dataset from the Cocoa Research Institute
of Nigeria, Ibadan (n=213) with a blocking factor having
4 levels.
The data were analyzed using exploratory data
analysis and the response surface methodology. The
exploratory data analysis relationship/distributional plot
shows that there exists significant negative relationship
between the yield of cocoa and the predictors
(temperature and rainfall). The estimated boxplot with
respect to blocking factor indicates that there is presence
of outlier in the yield of cocoa with majority of the yields
measured below 500kg over the period of study.
Results from the response surface models without
blocking indicate that all the estimated models were
statistically significant with all the lack of fit test
estimated to be insignificant (an indication of good fit).
On the basis of incorporated blocking factor to the
experiment, we observed that all models which range
from first order to second order outperformed those
without blocking factor by considering the estimated
adjusted R
2
. The blocking factors incorporated into the
experiment were found to be statistically significant with
all contour plots on the basis of the Eigen analysis
suggesting insignificant lack of fit. This implies that
incorporating blocking factor helped minimize the sum
of squared error and in turn improved the precision.This
study recommends that CRIN and other cocoa farmers
should learn to adopt newly developed techniques that
could militate against the impact of weather change
being experienced.
Keywords :
Response surface model, cocoa yield, rainfall, temperature, Eigen-analysis, contour, lack of fit.