Restoration of Missing Time-Series Data via Multiple Sine Functions Decomposition with Guangzhou-Temperature Application

The restoration of missing data is an important concern for data analysis. In this paper, an algorithmically innovative model termed multiple sine function decomposition (MSFD) model is proposed and developed for restoring the missing data about monthly average temperature (MAT) of Guangzhou, which is a representative major city of China. The proposed MSFD model is formed by successive approximation based on the existing data. After that, the MSFD model with parameters and structure determined is exploited to restore the missing data. Experimental results indicate that the proposed MSFD model can effectively estimate the intentionally removed data, and the values of the restored data are quite close to the values of the true data. In addition, with quantitative and qualitative analysis, the effectiveness of the proposed model is further illustrated.