Performance Augmentation of Base Classifiers Using Adaptive Boosting Framework for Medical Datasets
Table 1
Results for accuracy and percentage error of all the classifiers in AdaBoost framework for five medical (cancer) datasets.
Classifiers
Datasets
Leukaemia
Lymphoma-I
Lymphoma-II
GCM
Data set C
Instances
38
45
96
144
60
Attributes
7130
4027
4027
16064
7130
Naïve Bayes
Correctly classified
94.74%
91.11%
75.00%
16.67%
60.00%
Without AdaBoost
94.74%
91.11%
75.00%
16.67%
61.67%
Incorrectly classified
5.26%
8.89%
25.00%
83.33%
40.00%
Without AdaBoost
5.26%
8.89%
25.00%
83.33%
38.33%
Decision stump
Correctly classified
89.47%
86.67%
51.04%
16.67%
63.33%
Without AdaBoost
89.47%
82.22%
51.04%
16.67%
68.33%
Incorrectly classified
10.53%
13.33%
48.96%
83.33%
36.67%
Without AdaBoost
10.53%
17.77%
48.96%
83.33%
31.68%
Voted perceptron
Correctly classified
78.95%
97.78%
xx
16.67%
58.33%
Without AdaBoost
73.68%
84.44%
xx
16.67%
65.00%
Incorrectly classified
21.05%
2.22%
xx
83.33%
41.67%
Without AdaBoost
26.31%
15.55%
xx
83.33%
35.00%
Stacking
Correctly classified
71.05%
51.11%
47.92%
16.67%
65%
Without AdaBoost
71.05%
44.44%
47.92%
16.67%
65%
Incorrectly classified
28.95%
48.89%
52.08%
83.33%
35%
Without AdaBoost
28.95%
55.56%
52.08%
83.33%
35%
Bagging
Correctly classified
92.11%
93.33%
86.46%
xx
61.67%
Without AdaBoost
84.3%
86.67%
70.83%
xx
66.00%
Incorrectly classified
7.89%
6.67%
13.54%
xx
38.33%
Without AdaBoost
15.78%
13.33%
29.17%
xx
33.00%
J-48
Correctly classified
84.21%
82.22%
86.46%
xx
56.67%
Without AdaBoost
84.21%
77.78%
81.25%
xx
58.00%
Incorrectly classified
15.79%
17.78%
13.54%
xx
43.33%
Without AdaBoost
15.79%
22.22%
18.75%
xx
42.00%
Random tree
Correctly classified
81.57%
64.44%
66.67%
38.19%
68.33%
Without AdaBoost
86.84%
68.89%
58.33%
40%
63.33%
Incorrectly classified
18.42%
35.56%
33.33%
61.81%
31.67%
Without AdaBoost
13.15%
31.11%
41.66%
60%
36.68%
Random forest
Correctly classified
79.41%
91.11%
78.13%
52.08%
65.00%
Without AdaBoost
88.23%
91.11%
82.29%
50%
66.67%
Incorrectly classified
20.58%
8.89%
21.88%
47.92%
35.00%
Without AdaBoost
11.76
8.89%
17.7%
50%
33.33%
Bayes network
Correctly classified
94.74%
97.78%
90.62%
16.67%
62.34%
Without AdaBoost
94.74%
97.78%
90.83%
16.67%
68%
Incorrectly classified
6.66%
2.22%
9.375%
83.33%
37.36%
Without AdaBoost
6.66%
2.22%
8.16%
83.33%
31%
AdaBoost
Correctly classified
89.47%
86.67%
51.04%
16.67%
63.33%
Without AdaBoost
89.47%
86.67%
51.04%
16.67%
63.33%
Incorrectly classified
10.53%
13.33%
48.96%
83.33%
36.67%
Without AdaBoost
10.53%
13.33%
48.96%
83.33%
36.67%
ZeroR
Correctly classified
71.05%
44.44%
47.92%
16.67%
65%
Without AdaBoost
71.05%
44.44%
47.92%
16.67%
65%
Incorrectly classified
28.95%
55.56%
52.08%
83.33%
35%
Without AdaBoost
28.95%
55.56%
52.08%
83.33%
35%
Input mapped classifier
Correctly classified
71.05%
44.44%
47.92%
16.67%
65%
Without AdaBoost
71.05%
44.44%
47.92%
16.67%
65%
Incorrectly classified
28.95%
55.56%
52.08%
83.33%
35%
Without AdaBoost
28.95%
55.56%
52.08%
83.33%
35%
The crossed (xx) cells show that the results could not be generated for the specific classifier because of the limitation of the framework or the data set. Hence, the evaluation of these classifiers’ results has been carried out manually to check if any better results could be gathered for comparison. Bold values indicate the improved values.