Continuing problem 8 there is another method that has been suggested when the dependent variable has only a small percentage of one category in this case Y yes This is oversampling To do this you train the neural net on a subset that is more equally divided between Y and N values This allows the net to learn better about the Y cases because it sees a much higher percentage of them NeuralTools lets you do this fairly easily by creating a Tag variable as explained in the file P1715xlsx Then when y