Hospital foundation actions: Neural network model variable importance

The ability to model a set of complicated, non-linear relationships among variables is one feature for which neural networks are known. This research focused on using a neural network to uncover the best indicators of revenue generation in a specific type of nonprofit organization. Specifically, data from non-profit foundations supporting hospitals of various sizes in the United States was collected and analyzed in order to understand how revenue amounts are generated by nonprofit hospital foundations. The variable importance generated by the model indicates what variables contribute most to a foundation's yearly income. These results have implications for foundations in structuring their choices about how their foundation is run.