Parkinson's Disease (PD) is a progressive neurodegenerative disorder affecting motor functions, characterized by symptoms such as tremors, bradykinesia, rigidity, and postural instability. These symptoms stem from dopamine deficiency caused by neuronal degeneration in the substantia Nigra pars compacta. Early detection is crucial to improving patient outcomes and quality of life. This study presents a novel approach to detecting Parkinsonian symptoms using a wearable device integrated with machine learning algorithms. The proposed system features a custom wristband equipped with an Arduino UNO and Myoware 2.0 muscle sensor to capture electromyography (EMG) signals. These signals are normalized and analyzed using Fourier transforms to extract tremor frequencies, which are subsequently fed into machine learning models for classification. A dataset comprising 12 participants with four distinct tremor types was used to evaluate 30 machine learning and deep learning models developed in MATLAB. The subspace KNN model achieved the highest validation accuracy of 91.1%, demonstrating its potential for reliable real-time application. The system offers an affordable and non-invasive method for early PD diagnosis, providing a scalable solution for clinical and at-home use.