Obtaining users’ social attribute classification through their electricity consumption data is highly important for power systems to achieve personalized services. This paper proposes a 3 M-CNN-SC-SRU model based on deep learning, which realizes accurate classification of users’ social attributes by intelligently learning multiscale and multilevel features from users’ electricity consumption data. Experiments on the CER dataset verify the effectiveness of the proposed model. The results show that for the classification of P2 attributes, the proposed model improves the accuracy by an average of 15.34% and the F1 score by an average of 18.26% compared with existing models. For the classification of users’ social attributes, the accuracy of the proposed model improves by an average of 24.84%, and the F1 score improves by an average of 33.22%.