The Factor Analysis For Multivariate Time Series Of The Monthly Average Temperature Data

This paper is an application of the factor analysis for multivariate time series in meteorology.Firstly we perform a descriptive analysis for the monthly average temperature data of 8 cities in south China and its seasonal difference.It shows that the monthly average temperature series of these 8 cities and their seasonal differences may all have common factors.Therefore,we can fit the factor models for multivariate time series for the monthly average temperature data of these 8 cities and its seasonal difference respectively.According to the descriptive analysis of data,the paper uses the approximate factor model of Pan and Yao(2008)to fit the monthly average temperature data,and applies the approximate factor models of Lam and Yao(2012)to fit the seasonal difference data.The result shows that the monthly average temperature series of 8 cities in south China have five common factors,two of which are periodic factors.This is very similar to the monthly average temperature series of 7 cities in east China.They have four common factors,two of which are also periodic factors.In addition,the seasonal difference series of 8 cities have two common factors,which consist of one strong factor and one weak factor.