Comparison of Several Machine Learning Algorithms for Prediction

Abstract

Currently, the analysis of bacterial drug resistance is a research hotspot in the biomedical field. However, due to the long culture experiment cycle for traditional multi-antibiotics, the tolerance degree of a specific bacterium to a specific antibiotic cannot be quickly analyzed and accurately determined. Therefore, how to analyze and predict drug resistance is still a problem to be solved. With the development of whole genome sequencing (WGS) technology, this paper proposed a variety of machine learning methods to analyze and predict the drug resistance of Staphylococcus aureus and Neisseria gonorrhoeae. First of all, the original data was preprocessed; secondly, the key features, such as information entropy and Gini index, were extracted; finally, the drug resistance of each drug was analyzed and predicted. Experimental results showed that the training of this model could achieve rapid convergence, and the recognition accuracy was as high as 97%. By comparing the experimental results of various machine learning methods, it can effectively predict bacterial drug resistance.