Classification Performance Analysis of Decision Tree-Based Algorithms with Noisy Class Variable

Table 2

Accuracies for the C4.5/CART/RF algorithms on the data sets at varying noise levels.

Data set	0% noise	10% noise	20% noise	30% noise
Adult	86.19/84.45/86.64	85.82/84.46/85.98	85.39/84.34/84.68	83.47/83.17/81.00
Banknote Authentication	98.61/97.15/99.34	97.45/94.89/97.74	94.96/94.60/92.92	91.02/91.75/82.85
Blood Transfusion	78.19/78.78/74.73	77.26/76.47/75.67	77.00/76.99/70.72	75.26/75.00/67.23
Breast Cancer	95.14/94.35/96.81	92.56/94.57/95.85	93.00/94.14/93.71	92.13/91.99/89.27
Car	92.36/94.13/96.80	91.09/92.59/95.95	88.71/90.80/94.50	86.46/88.77/93.92
CMC	50.88/55.37/55.03	49.32/54.76/53.27	50.20/52.18/50.41	46.46/51.16/49.25
Congressional Voting	96.36/95.12/96.51	94.55/95.45/95.65	94.77/94.55/93.55	89.91/91.99/87.44
Dry Bean	91.13/87.06/92.44	90.18/87.13/92.15	88.41/87.03/91.65	83.37/86.41/90.90
Ionosphere	90.29/87.43/93.14	87.43/84.57/92.86	83.14/80.57/90.57	75.71/75.43/85.71
Iris	94.67/92.86/95.00	92.00/95.33/94.00	89.33/92.00/89.33	84.00/88.67/80.00
Optical Digits	89.87/76.91/98.09	87.83/75.13/98.04	84.16/74.11/98.12	75.42/72.51/97.83
Pen-Based Recognition	95.91/83.02/99.17	94.79/81.36/99.20	93.26/79.69/98.89	87.94/79.41/98.05
Qualitative Bankruptcy	98.00/98.33/100	98.40/98.00/99.20	97.60/98.40/94.40	95.20/92.40/84.00
Raisin	85.78/85.56/86.00	85.56/85.56/84.89	84.78/85.33/81.00	83.11/82.11/74.00
Seeds	91.90/91.50/93.50	87.62/90.00/91.43	84.76/90.48/90.00	80.48/87.62/83.33
Sonar	76.89/71.00/86.00	66.84/68.22/78.85	64.64/69.22/80.33	62.51/56.69/72.56
Tic-Tac-Toe Endgame	85.58/92.21/99.05	82.15/90.70/97.07	77.15/82.44/89.35	73.08/76.21/80.59
Vertebral Column	81.29/84.33/85.33	79.03/80.65/85.16	78.71/77.74/83.23	78.71/79.03/79.68
Wine	92.15/87.06/97.65	81.88/89.93/96.67	79.24/83.68/98.33	68.61/80.35/91.46
Wireless Indoor	97.30/96.55/98.40	96.20/96.35/98.35	95.70/96.00/97.75	88.40/95.45/97.00