As machine learning has greatly improved the depth and width of EEG analysis and psychological research, it has become possible to analyze the relatively stable personalities of individuals based on EEG. In this paper, we recorded the resting-state EEG of subjects, labeled them using their score of organizational commitment，and achieved automatic recognition. In order to complete such a new challenging classification task, we extracted a variety of different nonlinear dynamic features from resting-state EEG, and then used different machine learning models (SVM, GBDT, KNN, LR, Gaussian NB) to classify these features, and next evaluated experimental results based on cross-validation. The results show that Permutation Entropy and Approximate Entropy achieved best accuracy, which both obtained an overall accuracy of more than 70% based on machine learning. Furthermore, we adopted a stacking strategy and constructed a fusion model to improve the performance. The experimental results show that using the Stacking model to classify the Permutation Entropy of EEG can achieve an overall accuracy of 82.6% with 83.3% recall and 0.827 F1-score. In addition, we also conducted a comparative analysis of EEG signals with different lengths of sample and compare eyes-open and eyes-closed EEG.