Although the traditional fuzzy decision tree can extract fuzzy classification rules from fuzzy data, it can only obtain the membership information of nodes. It cannot obtain the non-membership and hesitation information of sample data for nodes, resulting in low accuracy of data classification. In order to solve this problem, a new Weighted Pythagorean Fuzzy Decision Tree (WPFDT) algorithm was proposed based on the Pythagorean fuzzy set theory. Firstly, the cluster center of continuous attribute data was obtained by the improved K-means clustering algorithm, and the continuous data was fuzzily processed by combining with triangular fuzzy numbers. Secondly,the weighted Pythagorean fuzzy entropy of each attribute was defined and calculated. The attribute with the lowest weighted Pythagorean fuzzy entropy was selected as the decision-making root node, and the attribute with the lowest fuzzy entropy was recursively selected as the splitting node under the root node. At the same time, the size of the tree was controlled by the threshold. The fuzzy rules of the path from the root node to the leaf node as well as the membership degree, non-membership degree and hesitation degree of the fuzzy rules were obtained, and the prediction classification was completed until the WPFDT model was generated. Finally, three medical data sets (Haberman, Breast Cancer, and Parkinson) on UCI were selected for the experiment, and the classification accuracy and the number of fuzzy rules were compared with three traditional decision tree algorithms (fuzzy ID3 algorithm, C4. 5 algorithm, and CART algorithm). The experimental results show that WPFDT is superior to other traditional decision tree algorithms in classification accuracy and tree size, and has higher recall rate and accuracy.