The incidence of diabetes is on the rise all over the globe. Therefore, a proper approach is necessary to identify the diabetic patients at the earliest and provide appropriate lifestyle intervention in preventing or postponing the onset of diabetes. Hyperglycemia and hypoglycemia are two important consequences of diabetes computed on the basis of blood glucose level. In this paper, we propose a machine learning approach to identify the probability of occurrence of hyperglycemia with the impact of physical activity (exercise). This prediction will be helpful in order to reduce the risk factor of hyperglycemia by timely taken preventive step and changing their lifestyle.