Based on existing researches, this paper discusses the model, which is the sum of randomvariable multiplication, and its statistic features in theory from continuous and discrete view, andgets unbiased results with measurable variance. In the condition of discrete, regression modelwith linear constraint is used to fit aggregate price index, and autoregressive distribute-lag modelis used to fit other price indexes. Without major technological revolution, construction cost modelsystem has been built, regarding aggregate as a constant. An index system, including aggregateprice index, artificial cost price index, material cost price index and machinery cost index), hasbeen set up, which is fitted to the model system. Take an example of national data, this paperseeks leading indicators and prediction model of all the price indexes. Artificial cost price index ispredicted by the leading variables of one and two years ahead, material cost price is predicted bythe leading variables of one and three quarters ahead, and machinery cost index is predicted bythe leading variables of three and four years ahead. |