Heart rate (HR) is a very important cardiological data that indicates a person’s health and effective status. Photoplethysmography (PPG) a promising approach to provide advanced and simple ways for estimating HR information as an unremarkable system on wearable devices. However, HR information may become unreliable due to motion artifacts during movement and physical activity, which lead to incorrect measurements. In this work, we analysis the performance of adaptive filter and machine learning (ML) algorithms to estimate HR using PPG signal. The method consisting of two main stages. In the first stage, adaptive motion artifact reduction consisting of three cascades Recursive Least Square (RLS) and cascades Normalized Least Mean Square (NLMS) adaptive filters is developed using three axis accelerometers signals as a reference, where cascade RLS and cascade NLMS are combined using convex combination scheme to further reduce the effect of motion artifacts. Stage 2 proposed ML based spectral tracking algorithms, whose aim is to locate the spectral peak corresponding to HR. In this stage, four different supervised ML algorithms (Support Vector Machine, Decision Tree, K- Nearest Neighbor and Logistic Regression) are Investigated to track the spectral peaks and the decision tree out performs all three algorithms with an accuracy of 98.96%. Experimental results on the PPG datasets including 23 subjects used in the 2015 IEEE signal processing cup showed that the proposed approach had a very good performance by achieving an average absolute error (AAE) of 1.98 beats per minute (BPM) and the personal correlation coefficient is 0.9899. AAE result proved that the proposed method provides accurate HR estimation performance in comparison with other existing works.