Abstract: Due to deep burial and the relatively dense structure of Q2 loess, its collapsibility is often overlooked. As a quantitative index to evaluate the degree of loess collapsibility, collapsibility coefficient is influenced by many factors, including soil moisture content, dry density, void ratio and other physical properties. Because there is a certain correlation among the factors, the correlation between the established collapsibility coefficient and the physical index is often of low accuracy. In order to effectively reduce the influence of multiple correlation of loess collapsible index on the data regression analysis results and to improve the prediction accuracy, the Q2 loess of the Weihua glycol project site in Binzhouis taken as the research object. Based on the statistical analysis of the correlation between the physical property index of the site stratum and the single index of collapsible coefficient and physical property, seven indexes with better correlation with collapsible coefficient are selected. By using principal component analysis and multiple linear regression analysis, the calculation model of Q2 loess collapsibility coefficient based on cumulative variance contribution rate is established. The comparison between the calculated value of the model and the measured value shows that the method is effective in reducing the multiple correlation and mutual influence between the influence factors of the collapsibility coefficient, and confirms the rationality and accuracy of the correlation between the established Q2 loess collapsibility coefficient and the independent influence factors.