Nowadays, the number of Thai elders is rapidly increasing among world elderly population, how to keep their health is a major concern. Cardiovascular Diseases (CVDs) which are severe diseases for Thai have higher mortality than cancers, and elderly people have a higher possibility to predispose CVDs. Hence, the risk factors for CVDs should be addressed. Obesity, as one of the risk factors of CVDs, seriously affects Thai elders' wellbeing; excessive sugar consumption is a way leading to overweight and obesity. The amount of consumed sugar by Thai is much higher than the standard sugar consumption, and it also could cause many other diseases. Therefore, this paper proposes a classification method for the elderly group who have the potential to control their blood sugar in order to prevent them from sugar overconsumption. This paper explored machine learning algorithms to find an appropriate classification method for elderly data. Artificial neuron network and K-nearest neighbors are applied for classifying elderly groups. Glycated hemoglobin (HbA1c) and fasting plasma glucose (FPG) are the noninvasive measurements of evaluating blood sugar, based on the two measurements, the 242 data from 121 elderly people are divided into two groups which are controllable group and uncontrollable group. The result indicates that the artificial neuron network is more suitable for the dataset with 70.59% accuracy as compared to the accuracy of K-nearest neighbors.