Weighted fusion algorithm for unbalanced power big data based on improved integrated learning

With the development of the society, the electricity consumption in each region increases dramatically, which brings serious challenges for data analysis and model construction due to the complexity and imbalance of electric power big data. Aiming at the problem of low classification performance of traditional methods in dealing with unbalanced power data, this paper proposes a weighted fusion algorithm for unbalanced power big data based on improved integrated learning. First, the existing integrated learning algorithm is optimized by combining the sample weight adjustment and the category distribution perception mechanism to improve the recognition ability of samples from a few categories. Second, a model fusion method based on the weighting strategy is designed, which improves the overall classification performance. Experimental results show that the algorithm proposed in this paper exhibits better precision, recall, and F1 scores on several real power big data sets compared with traditional methods, especially in minority class recognition.