The Application of Agglomerative Hierarchical Clustering for Obesity Classification

Obesity is the excessive accumulation of fat in the body which adversely affects the health and wellbeing
of the individual. It is a chronic and non-communicable disorder that poses socio-cultural,
psychological, clinical and public health challenges. The aim of this study is to apply Agglomerative
Hierarchical Clustering (AHC) to classify obesity and to develop a model employing Logistic
regression analysis for the prediction of obesity taking advantage of the relationship between Body
Mass Index (BMI), Age, Waist Circumference (WC), High-Density Lipoprotein (HDL)-cholesterol and
Low-Density Lipoprotein (LDL)-cholesterol. This Study was a work–site based cross sectional study
carried out on one hundred and twenty (120) workers at Judiciary Service Commission, Owerri, Imo
State, Nigeria. The Questionnaire was designed to address the background information of the
respondents with respect to gender, age, job title, department and address. The respondents were
anthropometrically examined and their lipid profile was estimated using the enzymatic colorimetric
method. Data were analysed using the Shapiro-wilks test of normality, Agglomerative Hierarchy
Cluster (AHC) analysis and Logistic regression analysis. These analyses were facilitated using
XLSTAT 2016 statistical tool. On the application of the Agglomerative Hierarchical Cluster Analysis
obesity was classified into Clusters 1, 2 and 3 with the majority of the obesed respondents being in
Cluster 1. The respondents in Cluster 1 belonged to the obesity class of overweight, while
respondents in Cluster 2 are of normal weight and finally respondents in Cluster 3 belonged to obese
class 1. A predictive model was developed based on Logistic regression analysis which showed a
strong positive correlation between obesity and HDL-cholesterol. The high profile of cardiovascular
risks identified in the study could be addressed through the provision of occupational health services
of which the ultimate goal should be the maintenance of urgent comprehensive health surveillance.