EMG Verilerinin Sınıflandırılmasında PCA Boyut İndirgeme Tekniğinin Etkisinin İncelenmesi Investigating the Effect of PCA Dimension Reduction Technique in Classifying EMG Data

This study aims to classify different hand movements using electromyography (EMG) data. The dataset used in the study is an 8-class problem. A completely balanced data set consisting of 1000 samples from each class is classified with different machine learning algorithms. 5 different machine learning algorithms are used to classify the data: K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Random Forest (RF), Naive Bayes (NB) and Logistic Regression (LR). To increase the efficiency of classification algorithms, dimension reduction is performed with Principal Component Analysis (PCA). In this way, the data is classified by reducing the original data to fewer than 8 attributes. In the classification performed with only 4 components, the highest performance is obtained from RF algorithm with 98% according to accuracy and F-measure metric. This result shows that PCA has a significant effect in classifying EMG signals.