This paper proposes Parkinson’s disease (PD) must be identified early in order to limit the illness’s course and improve access to drugs that slow its progression. Careful examination of the premotor stage is necessary to do this. Based on premotor characteristics, a machine-learning technique is presented to determine early on whether a person has Parkinson’s disease (PD). In particular, this study has considered several indicators, such as MDVP, Jitter, Shimmer, NHR, and HNR, in order to identify PD at the onset. The SVM approach demonstrates the higher rate of detection of the developed model using little data, which includes 147 early Parkinson patients and 48 healthy persons. Our approach makes clear how important some features are in the identification process as established by the SVM method.