Globally type 2 diabetes affects millions of individuals, a chronic illness that has a major negative impact on both health and economics. Early detection and accurate risk assessment are crucial for its effective prevention and management. Machine learning algorithms have been found to be useful in predicting and evaluating the likelihood of developing type 2 diabetes, as they provide valuable tools for risk assessment. In this study, a cohort retrospective strategy was used, involving a large and representative sample of individuals with and without T2DM. Feature engineering techniques were employed to extract relevant information from the dataset, and a thorough feature selection process was conducted to identify the most influential predictors. Data analysis plays a crucial role in healthcare systems. This research work investigates the use of various machine learning classifiers to forecast a person’s risk of developing type 2 diabetes. Here dataset included approximately 600 hundred patients record from hospitals in Bangladesh and is medically approved, which focuses on type 2 diabetes, and it consists diverse range of features, such as age, sex, sudden weight loss, polyuria, weakness, and polydipsia, collected from a cohort of patients. Various machine learning classifiers, including random forests, logistic regression, support vector machines, decision tree, and naive Bayes, have been employed to build predictive models. The results highlight the accuracy and other parameters of the different predictive models and their comparisons. Among all the models used in this paper, support vector machines (SVMs) give the best accuracy. Model with higher accuracy can predict almost precisely which patients are suffering from T2D or have a higher probability of developing diabetes in the future.