A parameter optimization method in predicting algorithms for smart living

Abstract

With the development of digital networks and communication, our life has become more intelligent and convenient. Automation systems and services are more and more popular with our daily life. To make full use of the data collected from smart living for accurate prediction, hybrid algorithms are proposed to optimize the parameters in machine learning. In this paper, multiple linear regression with a Kalman filter is proposed to forecast continuous variables, and support vector machine with a Kalman filter is proposed to forecast discrete variables. To present the effectiveness and accuracy of the proposed hybrid algorithms with less training data, we conduct experiments of predicting auction results based on two real datasets downloaded from a smarting system. The experimental results show that the proposed algorithms could improve prediction accuracy and reduce the error rates by calibration metrics.