Prediction of electrical power consumption in the household: fresh evidence from machine learning approach

Household power consumption assists the power supply department in determining how much energy people use and whether there are any unusual power consumption patterns. This study offers an extendable experimental analytical framework and analyzes the data in a visual manner using the household level electric power usage dataset as an example. Concerns about energy shortages and pollution have increased, highlighting the need to make full use of the limited electrical power accessible. The impact of multinomial regression, ridge regression, lasso regression, and polynomial regression on various features are compared in this experiment. In the experimental dataset, the effect of the polynomial regression model outperforms that of the multinomial regression, ridge regression, and lasso regression models. The necessity of establishing programmes to encourage the use of more thermally efficient locally available materials in building among other policy suggestions are provided. Household power consumption helps the power supply department understand the power consumption of residents and whether there will be some abnormal power consumption phenomena. Taking the individual household electric power consumption dataset as an example, this paper establishes an extensible experimental analysis framework and analyzes the data in a visual way. In the experiment, the effects of multinomial regression, ridge regression, polynomial regression, and lasso regression models on different characteristics are compared. The experiment shows that the effect of the polynomial regression model is better than the multinomial regression, ridge regression, and lasso regression model in the experimental dataset.