Disorders of consciousness (DOC) described by impaired wakefulness and awareness, can be categorized into Coma, Unresponsive Wakefulness Syndrome (UWS), and Minimally Conscious State (MCS). Resting-state EEG-based differentiation of these classes acts as a helping hand or even more to the conventional behavioral assessment methods in the diagnosis and prognosis of DOC patients. In this paper, multi-class classification of DOC patients using different machine learning models was performed and the results were analyzed using features like sample entropy, permutation entropy, and absolute and relative power extracted from resting state EEG data. The one-way ANOVA method determined the discriminative ability of the features with a post hoc Least Significant Difference (LSD) test. All four features showed significant differences (p < 0.05) in delta, alpha, and beta bands between the groups. The feature significance was also measured across the different brain regions as well. The classification results showed that the Random Forest classifier best classified the group with an accuracy of 78% and a precision of 88%.