Table 10 Comparison of explanations yielded by the single C4.5 tree and the descriptive forest

012

TP

ID 012 is predicted as “Heart Disease = Yes” because ST_Slope = Flat and Sex = M

TP

ID 012 is defined as “has CVD risks” from the voting by five main topics. The three main topics are independent: {Oldpeak > 0.85}, {ExerciseAngina = Y}, and {ST_Slope = Flat}. Two main topics, {ExerciseAngina = Y} and {ST_Slope = Flat}, are dependent main topics related to {Sex = M}, which promotes heart disease

Moreover, the risks of Instance 012 from each main topic can be described. For the {Oldpeak > 0.85}-tree, this instance describes the risk assumed by having MaxHR <  = 150 and Sex = M. For the {ExerciseAngina = Y}-tree, this instance describes the risk assumed by having MaxHR <  = 150. For the {ST_Slope = Flat}-tree, this instance describes the risk assumed by having Sex = M. For the {Sex = M, ExerciseAngina = Y}-tree, this instance describes the risk assumed by having MaxHR <  = 150. For the {Sex = M, ST_Slope = Flat}-tree, this instance describes the risk assumed by having ChestPainType = ATA, Cholesterol <  = 245, and MaxHR <  = 130

483

TP

ID 483 is predicted as “Heart Disease = Yes” because ST_Slope = Up, ExerciseAngina = No, FastingBS = Yes, and Oldpeak > 0.4

TP

ID 483 is defined as “has CVD risks” from the voting by two main topics, {Oldpeak > 0.85} and {Sex = M}. {Sex = M} is from a dependent tree that includes {ExerciseAngina ≠ Yes}, {ST_Slope ≠ Y}, and {ChestPainType ≠ ASY}

The risks of Instance 483 from each main topic can be described. For the {Oldpeak > 0.85}-tree, this instance describes the risk assumed by having MaxHR <  = 150 and Sex = M. For the {Sex = M}-tree, this instance describes the risk assumed by having ST_Slope = Up, Oldpeak > 0.4, and FastingBS = Yes