RETRACTED: Resident consumption expenditure forecast based on embedded system and machine learning

Abstract

The relationship between income and expenditure is significant for understanding the shape of the residents' economic dynamics. Consumer spending and household disposable income by the relationship between Machine Learning (ML) programming nonparametric locally weighted scatterplot smoothing regression analyzes. This study aims to determine the relationship between variables directly rather than the traditional parameters of regression. According to the survey, the usual assumptions, income and number of residents increase the impact of a sharp increase in spending at first, then slowed down. This growth can be relatively high mandatory spending little residents to explain. It increases according to income levels in middle-income and high-income groups. With the changes in population size, expenditure changes are limited in middle-income levels, restricted in most high-income levels. The latest growth in the smart meter housing sector is a large data set. Near real-time access to each household's power consumption by both supply and demand can extract valuable information for effective energy management. Predicted consumption will help improve power generation companies and demand-side management programs, but it is not. Consumption of individual households is very irregular, is a trivial task. In the business field of energy load forecasting, machine learning involves a lot of work. Machine learning methods are the so-called black-box model. The internal dynamic is almost unknown. However, without manual intervention, learning the ability to complex internal representation is a significant advantage.