Penerapan Linear Discriminant Analysis Untuk Meningkatkan Kinerja Algoritma Support Vector Machine

Obesity is a complex chronic disease influenced by various factors, such as genetic, environmental, and lifestyle, which is characterized by excess body weight due to the excessive accumulation of body fat. With the rapid advancement of technology and digitalization across all sectors, data has become increasingly vital, as large datasets generate valuable information. However, a key challenge in data analysis is addressing redundancy, noise, and high dimensionality, which can affect the performance of machine learning algorithms. This study aims to investigate the effectiveness of combining Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) in enhancing the accuracy and efficiency of high-dimensional data classification, particularly in predicting obesity levels. LDA is employed to reduce data dimensionality while retaining the most relevant features, whereas SVM is utilized as the classification algorithm to predict obesity levels based on patterns identified within the dataset. The research was conducted using a dataset consisting of 779 training samples and 195 testing samples. The results reveal that the combination of LDA and SVM achieved a classification accuracy of up to 99%, with a 50% reduction in data dimensionality and a computation speed of 0,0696 second. Moreover, computation time was significantly reduced, indicating that LDA not only facilitates data simplification but also improves the overall efficiency of the classification process.