Forecasting cocoa production of six major producers through ARIMA and grey models

Abstract

PurposeIn the current study, two grey prediction models, Even GM (1, 1) and Non-homogeneous discrete grey model (NDGM), and ARIMA models are deployed to forecast cocoa bean production of the six major cocoa-producing countries. Furthermore, relying on Relative Growth Rate (RGR) and Doubling Time (Dt), production growth is analyzed.Design/methodology/approachThe secondary data were extracted from the United Nations Food and Agricultural Organization (FAO) database. Grey forecasting models are applied using the data covering 2008 to 2017 as their performance on the small sample size is well-recognized. The models' performance was estimated through MAPE, MAE and RMSE.FindingsResults show the two grey models fell below 10% of MAPE confirming their high accuracy and forecasting performance against that of the ARIMA. Therefore, the suitability of grey models for the cocoa production forecast is established. Findings also revealed that cocoa production in Côte d'Ivoire, Cameroon, Ghana and Brazil is likely to experience a rise with a growth rate of 2.52, 2.49, 2.45 and 2.72% by 2030, respectively. However, Nigeria and Indonesia are likely to experience a decrease with a growth rate of 2.25 and 2.21%, respectively.Practical implicationsFor a sustainable cocoa industry, stakeholders should investigate the decline in production despite the implementation of advanced agricultural mechanization in cocoa farming, which goes further to put food security at risk.Originality/valueThe study presents a pioneering attempt of using grey forecasting models to predict cocoa production.

Demand Forecasting of Toyota Avanza Cars in Indonesia: Grey Systems Approach

“…Overall, the statistical models showed worst performance. The results compared through the MAPE test proved that EGM (1, 1, α, θ) is the best model for sales forecasting because the MAPE% values in both in sample and out of sample were <10 (high forecast accuracy) as the Lewis scale as well as literature (Quartey-Papafio et al, 2020) asserts that predictions with the MAPE value of less than 10% are very accurate. Based on the results shown in Table 3, it can be determined that the appropriate model for the demand of the Toyota Avanza is EGM (1,1, α, θ) because it has the lowest Mean Absolute percentage (MAPE) level, there are (4.2%) for MAPE in Sample and (5.9%) for MAPE out Sample.…”

Forecasting the Demand of Oil in Ghana: A Statistical Approach

“…Growth rate and doubling time analyses make forecasts more useful for decision-makers. The expression for the relative growth rate ( ) and doubling time ( ) are given by (Javed & Liu, 2018;Quartey-Papafio et al, 2020):…”

Development of the GM(1,1,⊗b) model with a trapezoidal possibility function and its application

“…Li and Wu and other researchers (Li et al, 2019(Li et al, , 2020Zeng et al, 2020c;Wu et al, 2018;Ma et al, 2019b) published some latest research results. (4) The combination of GMð1; 1Þ model with other elements and equations has resulted in many new models, and each new model has its particular advantages (Wang et al, 2017(Wang et al, , 2018Guefano.s et al, 2021;Quartey-Papafio et al, 2020;Ma et al, 2019c). For example, Wang and Ma combined the GMð1; 1Þmodel with the seasonal transformation and the Bernoulli equation to construct a new model for power consumption and income forecasting.…”