International Journal of Machine Learning and Cybernetics A Study of Repetitive Demand Prediction Based on Integrated Learning and Time Series

Abstract

Abstract In this paper, we propose a time-series-based method to analyze the process of ‘‘repetition‘‘. The method is able to obtain the repetition reliability of the detectee from the attribute information of the detection target and the detection index at different time points, and to predict the probability of the possible outcome of the next detection. We address the local optimum phenomenon that may be brought about by traditional time series due to the low relevance of data dimensions and optimize and improve on the classical time series analysis model. We also combine an integrated learning model for prediction after comprehensive processing of the data. In this paper, we take the problem of the repetition testing of AIDS patients in medical analysis as the background environment, and rely on the experimental results obtained from simulation experiments to outperform the experimental prediction results of the proposed model. An accuracy rate of about 91.8% was achieved. Thus, the feasibility of the method is verified. It can reduce the repetition detection rate and improve the detection efficiency to a certain extent. It also saves unnecessary testing costs.