Early Detection of Heart Disease on Imbalanced Data Using Gradient Boosting

heart disease is a major worldwide health issue due to its substantial impact on mortality rates. Prompt recognition is crucial for efficient management and containment. Emerging technologies, such as data mining and machine learning algorithms, can supplement traditional approaches in preventative endeavors. Nevertheless, the exceptional quality of the data can pose challenges when it comes to detecting and preventing cardiac disease in its first phases. This study aims to identify cardiac disease by employing a gradient-boosting technique on an imbalanced dataset. The employed approach involves the utilization of the resampling technique for data management. The resampling procedure used various strategies, including oversampling, under-over, under-over, and over-under sampling. The Gradient boosting algorithm was used five times, using four different resampling approaches and once without resampling. This was done to evaluate the effectiveness of each technique in handling uneven data and achieving superior results. The findings demonstrated that the amalgamation of gradient boosting with the over-under resampling technique yielded a notable degree of precision, suggesting its promise in the timely detection of heart disease. The percentage of values below is over 14 percent higher without sampling. The research has significant implications for data mining in the health sector and provides suggestions for future research.