This study constructs an emotion identification system based on a valence/arousal paradigm using electroencephalography (EEG) signals. Logistic regression is used for feature selection with correlation coefficient score applied to guarantee that such a model does not overload the data, randomized divides among all occurrences are performed. The performance of emotion recognition systems improves caused by means of brain waves is determined by the methodologies used to features extracted, the selection of features, and the classification process. Inside this paper, several machine learning techniques are used viz. Multilayer Perceptron, Logistic Regression, Random Forest, Gaussian Naïve Bayes, Decision Tree, Support Vector Machine, and ensemble techniques with voting classifiers. To complete the build and creation of an expert the system for automatic detection of emotions from EEG data, this kind of comparison research utilizing machine learning must be laid out on this subject. The experiment has been carried out for all the Machine Learning techniques using all attributes. Experiments have demonstrated the efficacy of the proposed method with 98.45% accuracy for Decision trees and random forest Classifiers.