2. Build a real-world neural network. Using demo software downloaded from the Web (e.g., NeuroSolutions at neurodimension.com or another site), identify realworld data (e.g., start searching on the Web at ics.uci.edu/

~mlearn/MLRepository.html or use data from an organization with which someone in your group has a contact) and build a neural network to make predictions.

Topics might include sales forecasts, predicting success in an academic program (e.g., predict GPA from high school rating and SAT scores; being careful to look out for “bad”

data, such as GPAs of 0.0), or housing prices; or survey the class for weight, gender, and height and try to predict height based on the other two factors. You could also use U.S. Census data on this book’s Web site or at census

.gov, by state, to identify a relationship between education level and income. How good are your predictions?

Compare the results to predictions generated using standard statistical methods (regression). Which method is better? How could your system be embedded in a DSS for real decision making?