Research On Bayesian Network Model Of Performance Prediction Of Enterprise Innovative Project Portfolio

For maintaining long-term competitive advantage,enterprises often carry out parallel multiple innovation projects to maximize the use of resources,but also makes the importance of innovative project portfolio management gradually highlighted.The success of the innovative project portfolio can be presented through innovative project portfolio management to the extent to which the project portfolio's strategic objectives are completed,ie,innovative project portfolio performance.Ensuring higher innovative portfolio performance requires the control of the entire process of innovative project portfolio management and the timely detection of key factors that affect success.Project portfolios' success is influenced by top management involvement,project managers' competency and the process of portfolio management.There exist not only linear but also nonlinear causal relationships among these influential factors.However,only linear relationships are simulated in traditional methods for performance prediction.In this paper,a Bayesian network model with latent variables is proposed and a framework for predicting R&D portfolio performance is proposed to solve this problem and enhance the accuracy of prediction.The influential factors of R&D portfolio performance are identified by questionnaire and the traditional Bayesian networks(BN)structure for prediction is developed firstly;then structural equation models are developed based on the hypothesized causal relationships among the latent variables in the BN structure.Partial least squares(PLS)are used to test the false relationships in the models.Lastly,a BN model with the latent variables is constructed and the R&D project portfolio performance is predicted through the parameter learning of the model.In addition,the method proposed in this paper is applied to the collection of 169 enterprise sample data,and the ten-fold cross method is used to compute the average accuracy of innovative project portfolio performance predicted by the network structure of the artificial neural network,the network structure of the tree-extend Bayesian classifier learning,the network structure of the K2 algorithm learning and the network structure proposed in this paper.The results show the network structure proposed in this have high accuracy and strong stability.