CS 4804 Homework #5

1. (100 points) Consider the "heart disease" dataset from the UCI machine learing repository. Use only the Cleveland portion of this dataset (read the ".names" file) and only the 14 attributes out of the entire 76 attributes (again, read the ".names" file). Use the "processed" version of the Cleveland dataset, not the raw version. Divide the 303 entries in this dataset into training, test, and validation sets, taking care to ensure that the distribution of classes is approximately the same across the new datasets (note that there are five classes).
   
   Design a neural network to learn to classify patients. For full credit, explain the choice of the number of neurons you use in each layer, how you encoded the data as inputs/outputs to your network, plots of performance as the network is trained, and final performance on the test set. Make qualitative observations about the performance of your neural network and compare it to information in the ".names" file.