The activity of nano Pt-Ru catalysts is a multivariate function of particle size, alloyed degree, oxide composition, and so forth. The monodependencies of catalytic activity on individual structure parameters (structure-activity relationship, SAR) are of great importance but, unfortunately, unobtainable in practical measurements. A pattern-recognition methodology is proposed for the first time to extract SAR information from all of the relative experimental data, which we hope will cast new light on the in-depth understanding of this important catalyst. As a preliminary demonstration, a multivariate linear regression and a generalized regression neural network were applied to analyze a small data set for methanol oxidation. It was found that both increasing the content of hydrous ruthenium oxides and decreasing the particle size would benefit the catalytic activity, whereas the effect of the Pt-Ru alloy degree turned out to be unremarkable.