A comparison of neural networks with regression techniques

Regression techniques are widely-used in a variety of problems for classifying or forecasting outcomes. However, a new technique based upon neural networks may be superior in many cases. Here, we show how neural networks can be used to forecast more accurately than with regression techniques. In a comparison of neural networks with regression in seven problems, the neural network accuracies were greater in six of the problems and the same as the regression accuracy in one of the problems. Further, neural networks require no assumptions about the data being modeled (as regression techniques require).