A severe mental disorder, schizophrenia is characterized by changed feelings, thoughts, and behavior. Delusions, hallucinations, and social disengagement are some of the symptoms of this disorder. As diagnosis is mostly based on the interviews by psychiatrist, there is a chance of human error. This emphasizes the need for novel techniques like machine learning applied to EEG signals. Enhancing the accuracy and efficacy of schizophrenia diagnosis could be achievable with the use of machine learning algorithms to EEG data. Timely treatment can significantly enhance the quality of life and outcomes for persons diagnosed with schizophrenia, but early detection is essential. This study presents a novel variable mode decomposition (VMD) approach for EEG signal classification. Variational Mode Functions (VMFs), which indicate various frequency ranges, are created by breaking down EEG data from nine channels. Nine statistical features are taken out of each VMF, giving each channel 45 features. These attributes are subsequently classified using various machine learning algorithms. The suggested method’s performance is assessed by evaluating the classification accuracy for each of the five VMFs and all nine channels. The best accuracy is given by K-nearest neighbor (KNN) which is 91.57%.