A PLS-SVM Model for Efficiency Evaluation of Libraries

Beneficial efficiency evaluation of libraries is to analyze and study the administration and demands for information sources. In order to solve address problems of fewer samples for evaluation and the high correlation between evaluation indicators, a partial least squares-support vector machine (PLS-SVM) model of library efficiency is proposed. The model combines the advantages of both partial least squares and support vector machines, which not only possesses strong ability of overall optimization, but effectively avoids the linear relationship between variables by eliminating the correlation between indicators to improve the accuracy of evaluation. Compared with the simple support vector machine results and the problem of discriminate analysis results, the results of model report from 14 colleges and universities have shown that the predicted accuracy of beneficial efficiency evaluation of libraries can achieve 100%. It has shown that PLS-SVM model can evaluate the efficiency of the library accurately with results consistent with reality.