The Analysis Of CPI Trend And Its Influencing Factors In China Based On SVR

Inflation is an important economic issues related with the economic development and people's lives.As a widely used price index in the world, the consumer price index has been not only one of the inflation targets, but also the important reference of the formulation and implementation of policy.2007-2008,China experienced a price increase,unlike any previous ones,and its affecting factors are more complicated.There are both cost-push factors and stimulating demand factors,for the impact of both domestic and international,and affected by the total reasons while structural reasons.Therefore,concerning and judging the trend of the consumer index,as well as analysis of various economic and financial variables affecting the price of the direction and extent,has become an evasive and urgent problem.Taking into account the non-stationary of the economic and financial sequences and the integrity and complexity of the economic system,This paper introduces support vector machines,a considerably advantageous statistical learning algorithm,uses SVR methods based on the autoregressive distributed lag,and establishes the basic SVR model, adaptive SVMs model and principal component analysis ASVMs model to analysis the monthly CPI trend in 1997-2008.And this paper applies saliency analysis to observe the different influence of factors during long-term and short-term in china,and uses factors changing analysis to analysis the possible change of the price trend cause by the factors' changing in the future.This paper is a successful application in China's consumer prices using SVR,and we obtained desired results and tested the superiority of SVR.This paper obtain the major conclusions as following:1.food prices, international prices and foreign exchange reserves are the main factors affecting prices in long-term,but recently,fixed investment,the stock market and other asset prices become more significantly effecting CPI, and the impact of external shocks is beyond the CPi(-1).2.The ASVMs based on KPCA model is quite suitable for China's price analysis during 97-08.Under the conditions of frequently external uncertainties,the short-term forecast has a better performance than long-term forecast,and the out-sample forecast of CPI in June 2008 is:107.26,the interval of a five percent margin error is[106.72,107.80].3.While in the near future, fixed investment,international prices shall remain significantly impact on prices,but the impact of PPI and domestic housing prices is not yet clear at present.Analysis also shows that China's macroeconomic policy has started to show their expected results.