PRINCIPAL COMPONENT ANALYSIS AND CLASSIFICATION OF SUGARCANE PRODUCTION DATA IN BURUNDI

Abstract

This study aims at identifying groupings of individuals (years) from sugarcane production data in Burundi by applying Principal Component Analysis (PCA) and classification, and computing clusters’ distances between centroids. These data were obtained from the Ministry of Environment, Agriculture and Livestock in Burundi and cover a period from 2000 to 2019, i.e. 20 years. The variables used are mean sugarcane yield, sugar production, cultivated area, sugarcane production, molasses production, herbicide inputs, mean temperature and cost of sugar production. R software version 3.6.1 was used to analyze data. This study shows that the years 2002 and 2013 are opposed by the first factorial axis while the years 2001 and 2017 are opposed by the second one. The years 2000, 2002, 2008, 2012, 2013 and 2017 are better represented on the principal factorial plane. The years 2017 and 2019 are characterized by high values of herbicide inputs and cultivated area while the years 2011, 2012 and 2013 are characterized by high values of mean temperature and mean sugarcane yield. The year 2015 is characterized by high values of sugar production, sugarcane production, cost of sugar production and molasses production.