Resumen
This paper looks at revenue amounts generated by non-profit hospital foundations throughout the US. A number of inputs, including, among others, compensation, type of support given to the hospital, type of foundation expenditures, and hospital size, were used to develop models of foundation revenue. Both neural network and regression models were developed and compared in order to see which one gave a better model and to see how they ranked the relative value of the input variables. Though the generated value of revenue for both models correlates highly with actual revenue, the neural network shows smaller error. The order of variable importance for the models is very different. Each model would have different implications for foundations in planning their next round of revenue generating events.