Research On The Unconsistent Results Of Multi-Model Of Comprehensive Evaluation On Venture Capital Projects

This paper focuses on the evaluation of the venture capital project and aims at improving the capability to evaluate the project of venture capital. Based on the detailed analysis of the characteristics, contents and previous research of venture capital projects, we concluded that comprehensive evaluation method is the most efficient way to evaluate the project of venture capital. In addition, we investigate the inconsistency existing in the results of comprehensive evaluation.Firstly, a comprehensive index system to evaluate the venture capital project is introduced. Considering that the project of venture capital is a multi-objective, multi-level system with complex structures and numerous factors, comprehensive evaluation method is applied to build the synthetically fuzzy evaluation model, gray correlation degree model, entropy model, principle component analysis.Secondly, due to the difference of different models, the results of evaluation may be inconsistent. We perform the preliminary research on the causes of sensitivity of evaluation results. The difference of the evaluation method mechanism , weighting factors, and manner of choosing index may be the main causes of inconsistency.Thirdly, after discussing the applicability and consistency of different methods, we put forward combinational evaluation which can resolve the inconsistency of the results. Combing the two behaviors of the comprehensive evaluation results, combinational evaluation based on nonlinear programming and genetic algorithm are built. Also, the results of combinational evaluation are tested. Finally, according to the methods and models we introduce above, six software projects that may be invested by venture capital are evaluated and the evaluation results rank from the best to the worst. This example demonstrates the applicability of our methods.