A conceptual neural model for business selection in multi business unit firms

Despite of its importance in the context of corporate portfolio management (CPM), business selection in multi business unit Arms is still among ambiguous problems because of its sophisticated nature in which the effective decision rules are different from one firm, industry or country to another. Therefore this article tries to address the business selection problem in corporations by applying a two layer neural networks with sigmoid activation function as a pattern recognition supervised learning algorithm to recognize business selection patterns in different Arms and make decision for adding or divesting new business units based on a set of quantitative features. The model trained, validate and tested and results were indicative that the model is relatively reliable for using in action.