Analysis of Net Primary Productivity (NPP) driving factors can provide a scientific basis for the ecological environment monitoring and prediction, and assessment of ecological carrying capacity. The NPP drivers' research has become a hot topic in recent years, but previous studies failed to reflect the spatial attributes of factor in the original data and analysis. Actually, it was unfavorable with achieving more scientific consequence, because it put a brake on expression of the spatial information of the data itself. Therefore, we selected the Xinjiang Yili Valley as our study area and analyzed the environmental drive factors of semi-arid region of NPP by attempting to use a new spatial pre-treated method (C.V computation, C.V=SD/Mean) for data, which can help to express the spatial information of the ecological data itself. By a comparative analysis between normal data sets results and C.V data set results, we obtained some conclusions:(1) in semi-arid region, NDVI, accumulated temperature, and altitude were the most drive factors of NPP (P<0.01). It is mainly reflected in the positive promoting effect of the accumulated temperature on vegetation growth and the control of elevation change to the local precipitation condition. We also found that these factors manifested superiority on the importance rankings. (2) The C.V data set had higher fitting degree (P<0.01) in the model establishment, and had a certain application prospect. The fitting degree of the ordinary data sets is about 0.10 lower than that of C.V data sets. (3) After the original image was processed by C.V pre-treated method, it could directly express the spatial fluctuation of the factor, and the C.V computing process highlighted the fluctuation of the data between small and nearby domains. (4) After Duncan analysis, we found that each factor had significant differentiation characteristics at elevation level (P<0.01). Furthermore, the results confirmed the effect of altitude factor on NPP. From all above, C.V computing process allows the original data set to have spatial attributes. The statistical analysis results are not dependent on the numerical relationship, the relationship of our results can express more comprehensively. This analysis process and results have important experimental value and the scientific significance for the analysis of ecological environmental factors and the driving force analysis of NPP.