The study level of karst collapses is still macroscopic and qualitative right now. A lot of work about the hazard forecast methods of Karst collapses has been done, the application of many new methods such as grey statistics, neural network, and AHP-fuzziness compositive evaluation has heightened the precision obviously and a lot of better effictiveness has been obtained. The PLS Path Model adopt the iteration solution of monogenesis or multivariant linear regression, this model does not need hypothesizing the probability distribution of observation variables and the requirement of sample point capacitance is very loose, and the problem that the model...