Mental stress is currently a significant concern, especially among the young. Stress adversely affects the overall performance of people’s work, and in certain cases, can even cause serious health issues. Everyone experiences stress in life. A unique way to identify and classify stress levels based on Electroencephalogram (EEG) is proposed in this manuscript. In this work, fast Walsh Hadamard transform is used to generate all frequencies which exist in the EEG signals. The range of alpha, beta, gamma, and delta from index value is calculated in subsequent stage. Principal component analysis (PCA) is applied for the feature dimensional reduction which is followed by the standard scaler. The PSD vector has been calculated for healthy and unhealthy EEG signal groups using the Welch method. The PSD vector is used an input to the voting classifier which is the combination of the k-NN and logistic regression classifier. The experimental results found that the proposed method provides better results when compared to the existing methods in terms of Accuracy (Acc) and Mean Square Error (MSE). The proposed method achieves a highest classification accuracy of 94.22%