Atrial Fibrillation (AF) is quivering or irregular heartbeat by which the two upper chambers of the heart (atria) gets affected resulting in disruption of blood flow throughout the body. Although AF itself usually isn’t deadly, it is a serious medical condition that sometimes requires emergency treatment. If left untreated, AF can be life-threatening leading to blood clots, stroke, heart failure and other heart-related complications which are the main causes of disabilities and deaths worldwide. In this research, we have tried to find an optimized algorithm to automatically detect normal and AF patient from ECG signal efficiently using ?CPSC-18 Challenge Data?. We have implemented SVM, KNN, and PCA-DA. The highest accuracy among these models is achieved by PCA-DA as high as 97%. The proposed high accuracy, low false alarm model for detecting AF has potential applications in wearable devices and can assist medical practitioners in decision making while removing cognitive biases.