Dam horizontal displacement prediction method

The invention discloses a dam horizontal displacement prediction method. According to the method, firstly, high-dimensional factors are subjected to dimension reduction and correlation elimination based on typical cause analysis, and linear transformation vectors for overall prediction of two kinds of horizontal displacement, namely from the upstream to the downstream of a dam and from the left bank to the right bank of the dam, are obtained; secondly, samples are classified according to condition similarity based on cluster analysis, singular points are screened out, and classified regression and prediction are conducted on the basis of classification. In this way, noise interference and information loss caused by unbalance of sample quality and sampling can be avoided, sample information can be excavated more reasonably and sufficiently, a prediction result with a higher confidence level can be obtained, and a more practical decision-making assisting function is achieved.