In recent years, fine dust has been closely related to everyday life due to the surge of domestic fine dust, and thus interest of people is increasing. Possible sources of fine dust include burning fuel, exhaust gas from factories and automobiles. In addition to these trends, many studies have been carried out to investigate the relationship between fine dust and related factors. In this study, the relationship between the air pollutants and weather factors was investigated.
On the other hand, as the machinery industry develops, a large amount of data is being produced, and methods for handling the data effectively are important. Among them, High-dimensional data is a much smaller number of variables than observations, and it is difficult to apply linear regression analysis method. In order to solve this problem, a penalized regression method using constraint condition is proposed. In this paper, various penalized regression methods are applied and the effects are compared. Also, by using the smoothing effect of the penalty point regression method, the interpretation was improved by finding a global meaning through smoothing using the neighboring region information.
As a result, the concentration of fine dust in Seoul was influenced by air pollutants, especially the concentration of sulfur dioxide (SO2) 6 days before and 13 days before. In addition, among the weather factors, we could detect the effect of the wind blowing from the west, and we could infer that the concentration of fine dust increases when wind is blown from China.
In case of spatial visualization of the nationwide fine dust concentration, the characteristics of the fused Lasso were used and smoothed to a similar value by giving a penalty to the neighboring neighbor value through the tuning parameter. According to the examples presented in this paper, natural irregular concentration patterns are complemented and natural interpretation is derived.