A Fitting Method of Dose-Effect Data of Traditional Chinese Medicine Fusing Softmax Regression and an Improved PSO Algorithm

Aiming at the problem that the dose-effect data of traditional Chinese medicine are of multi-dimensional structure, this paper introduces a kind of data fitting method which is relatively accurate, efficient and effective. This method utilizes softmax regression function to build a fitting model for the multidimensional dose-effect data of traditional Chinese medicine and this paper suggests an improved particle swarm optimization (PSO) algorithm to make the fitting of the multi-dimensional dose-effect data of traditional Chinese medicine. Firstly, this paper adopts Min-Max normalization method to normalize the data. Secondly, this paper utilizes a transformation matrix to make a fast dimensional transformation. Thirdly, this paper uses softmax regression function to build the mathematical data fitting model and designs a kind of softmax evaluation function for computing adaptability in PSO algorithm. In the end, this paper extends PSO for single variable to the optimization of multidimensional variables, designs a kind of simplified velocity formula, adopts a kind of gradual learning and update strategy, and applies the improved PSO algorithm to make the fitting optimization. From the results of the experiment, our method is more accurate than the other 2 methods and it's more efficient than hill climbing algorithm.