The use of electroencephalography (EEG) to identify mental stress is suggested by this study. High precision, sensitivity, and specificity rates were attained employing the original thesis’s effective strategy of employing Horthy features that were taken from a 32-channel dataset and categorized with the K-Nearest Neighbors (KNN) classifier. But because of the tiny sample size, more investigation using a bigger dataset was required for confirmation. A smaller, 8-channel EEG dataset was used for this investigation, and it had been cleaned and filtered similarly to the first project. However, the categorization findings were not as good as expected, thus other approaches were investigated. Accompanying conventional classifiers, Convolutional Neural Networks (CNNs), specifically EEG Nets, have been utilized to examine different feature harvesting and filtering strategies. In addition to acceptable performance generated by the Deep and Shallow CNN models, the greatest results utilizing classical classifiers have been achieved employing the whole RAW data, time-varying features of RAW data, and wavelet scattering characteristics of RAW data utilizing Support Vector Machine (SVM). Independent Component Analysis (ICA) was used for filtering, although it was not as successful as it was in the previous study. Particularly, the raw data produced encouraging results when it came to identifying mental stress when utilizing CNNs, indicating that greater care should be taken in future studies.