Parkinson's disease is a chronic neurological disorder that progressively causes tremors of the hand, stiffness, and slow movement of the feet. No medical cure is available yet, but early detection and proper treatment can help patients manage these conditions. Typically, medical practitioners examine patients manually and prescribe treatment plans based on their locomotive and neural symptoms observed in the medical room. However, consideration of the patient's performance during day-to-day activity may improve detection. This study uses machine learning models to help automatically detect this disease based on daily movement data. Accelerometer sensor signals captured from 34 participants' chests, arms, and thighs as part of the full PD-BioStampRC21 public dataset were used to train a machine-learning model that can effectively detect Parkinson's disease. The study obtained 92.6% accuracy for holdout and 94.4% for k-fold cross-validation. This proposed model will help medical practitioners with the early detection of Parkinson's disease and provide a pathway for treatment at an early stage,