| Summary: | The issue of obesity becomes more worrisome since it is ranked higher for people diagnosed with obesity and overweight. Therefore, this study aimed to identify the best predictive model amongst logistic regression, decision tree and artificial neural network models to predict overweight and obese senior citizens in Selangor, Malaysia. Data were collected amongst Malaysian senior citizens in Selangor aged 60 years old and above. Upon analysis, the decision tree, logistic regression, and artificial neural network predictive models delivered accuracy rates of 65.44%, 63.9%, and 63.71% respectively. The decision tree was selected as the best predictive model as compared to other data mining techniques. However, it is noteworthy that the accuracy of these findings was somewhat restrained, attributed to the constrained deployment of a wider array of variables. In addition, seven attributes were found to be the most important factors in predicting a person's weight status. These include cigarette smoking, age, presence of diabetes mellitus, frequency of light physical exertion, history of cardiovascular disease, dyslipidemia, and household income, each playing a crucial role in predicting an individual's weight status. © 2024 Author(s).
|
|---|