As a common yet chronic disease, Diabetes Mellitus (DM) affects millions of people all over the world. Some groups are more vulnerable to diabetes in comparison to others, such as people with a family history, and those suffering from obesity. Early detection of such people, in conjunction with preventive measures, can go a long way in saving the lives of people who are most likely to be infected with these diseases and avoid suffering. Consequently, Prediction solutions have been found using data mining techniques which help to discover hidden information about the disease and supports decision-making. This study aims to ensure that diabetes is predicated at the initial stages using two algorithms of machine learning: Random Forest and Multi-Layer Perceptron (MLP) using the WEKA environment to estimate the accuracy. The experiments were applied using a dataset obtained from the Machine Learning Repository of UCI. The dataset comprises 16 attributes and 520 instances collected via questionnaires from patients at Sylhet-based Sylhet Diabetes Hospital (Bangladesh). According to the findings, The RF method has been found to yield better results than MLP when it comes to enabling early prediction of diabetes with a high accuracy of 97.88%.