Yangtze River Delta is a very important region for China's economic development has played a pivotal role.Then their development mechanism of it is very necessary, but for regional development data, since the existence of space and space-related since the heterogeneous nature of the use of the general regression equation does not apply. In this paper, in recent years to develop the new GWR regression model of the Yangtze River Delta in 1997 and 2003 two years of regional development data were studied. The results show that different parts of the Yangtze River Delta on the same investment performance of the different elements is a reflection of this note for regional development in terms of cross-sectional data, GWR regression model is superior to ordinary least squares regression model, can be more For accurate results. In addition , according to results of the analysis of GWR, the Yangtze River Delta can be roughly divided into five regions, each region of the same elements of the sensitivity of inputs are not the same, reflecting the different geographical characteristics.