| Popis: |
The Hidden Naive Bayes (HNB) classification algorithm is a kind of structurally extended Naive Bayesian classification algorithm, which introduces a hidden parent node for each attribute so that the dependencies between attributes are utilized. However, in the classification process, the effect of the attribute pair on the attribute is ignored. Therefore, the double-layer Hidden Naive Bayes (DHNB) classification algorithm fully considers the dependence between attribute pairs and the attributes. However, he did not consider the contribution of different values of each feature attribute to the classification. To solve this problem, an improved DHNB algorithm was obtained by constructing a corresponding weighting function to calculate the contribution degree of each feature attribute value to the classification and using the obtained weighting function to weight the formula in the DHNB algorithm. Finally, the improved algorithm was simulated experiment on the University of California Irvine (UCI). The results show that the improved algorithm has higher classification efficiency than the original DHNB algorithm, and the method has good applicability. |
