Filter Model Research of Characteristic Value of Typical Construction Engineering Based on T Test and Decision Tree Method

It is necessary to collect a large number of engineering data in the compilation of construction cost index. It would be ideal if all the characteristics items of different engineering projects can be selected as the research base. But there is currently no uniformed standard of selecting the typical construction engineering characteristic values. Engineers tend to judge by the experience of the cost engineers. This causes low efficiency and lack of quantitative analysis. The article collects the characteristic value of 100 building constructions in the city of Wuhan during the period of October to December of 2016 as samples. It first analyzes the characteristic feature attributes by t test, and then filters the samples by feature classifying, adopting the decision tree method. The article also solves the overly-fitting problem and the multi-valued bias of the decision tree method so that the stability of the feature selection can be ensured. Lastly, the article builds a decision tree calculation model and verifies the accuracy of the model by SPSS software to achieve the goal of accurately selecting the characteristic value of typical construction projects.