Neural Network Predictive Modeling

Predictive Neural Network (NN) is a very powerful predictive modeling technique. It is derived from nerve systems (e.g., human brains). The heart of the technique is neuralnet (or network for short). A typical network consists of layers of nodes and links between neighboring layers' nodes. The first layer is input layer. Nodes of input layer represent induction fields or values of nominal induction fields. The last layer is output layer. Nodes of output layer represent prediction values (or class names), i.e., values of a target field. The rest of layers are called hidden layers (or middle or internal layers). There may be zero or more hidden layers, normally a single hidden layer. For example, the figure shown at the left-hand side contains a network that determines credit risk levels based on gender, age and salary. It consists of an input layer of 8 nodes, one hidden layer of 8 nodes and an output layer of 2 nodes.