SVM Classification Based Early Warning Method of Brain Drain

Talents are the most important resource of high-tech enterprises. Thus conducting an effective early-warning of the brain drain in high-tech enterprises, will effectively reduce the brain drain acts to reduce the loss of high-tech enterprises. This paper, using of high-tech enterprise day-to-day performance appraisal data, in accordance with the characteristics of Chinese high-tech enterprises, carry out combination Factor Analysis and SVM to establish an early warning method of the brain drain. First of all, standardize the original data, and then using of Factor Analysis to eliminate redundant data and extract feature vector, finally, through the experiment analysis of parameters and the adjustment of the impact of Kernel selection for support vector machine, and search for the optimal support vector machine model, through Matlab6.5 in the use of SVM evaluation the brain drain action of high-tech enterprises. Experimental results show that SVM suite to the actual of small samples of Chinese high-tech enterprises, the model can act effectively to identify the brain drain of high-tech enterprises, and has a good early-warning effect.