The Heart Disease dataset was donated to the Irvine collection by David Aha. The principal medical investigator was Robert Detrano, of the Cleveland Clinic Foundation. Table 17 shows the test costs for the Heart Disease dataset. A nominal cost of $1.00 was assigned to the first four tests. The tests in group A are blood tests that are thought to be relevant for heart disease. These tests share the common cost of $2.10 for collecting blood. The tests in groups B and C involve measurements of the heart during exercise. A nominal cost of $1.00 was assigned for tests after the first test in each of these groups. The class variable has the values "buff" (healthy) and "sick". There was a fifteenth column, which specified the class variable as "H" (healthy), "S1", "S2", "S3", or "S4" (four different types of "sick"), but we deleted this column. Table 18 shows the classification cost matrix. There are 303 cases in this dataset. We deleted all cases for which there were missing values. This reduced the dataset to 296 cases. In our ten random splits, the training sets had 197 cases and the testing sets had 99 cases.
