As machine learning technology advances quickly, more and more people pay attention to regression problems, which aid in extracting the law from vast amounts of data in order to produce a prediction effect. People now heavily rely on data prediction in their daily lives. The technology is currently widely employed in a variety of industries, including forecasting the weather, diagnosing illnesses, and the financial markets. As a result, one of the most active areas of research in the field of machine learning in recent years has been the study of machine learning algorithms in regression situations. This research mainly uses three popular machine learning algorithms: neural network, extreme learning machine, and support vector machine in order to more thoroughly explore the application effect of machine learning method in regression problem. The benefits and drawbacks of each machine learning method are then examined by contrasting the results of the single model and integrated model applications of various machine learning algorithms to regression situations. Here, we can consider diabetes to be a serious issue and then attempt to use a number of ML algorithms and build an ensemble model to compare the effectiveness of accurate diabetes prediction at an early stage. Here, we analyze several algorithms to determine which will provide the highest level of accuracy while taking into account a number of performance evaluation criteria.