Research Article

A Feature Selection Algorithm Integrating Maximum Classification Information and Minimum Interaction Feature Dependency Information

Table 3

Average classification accuracy (%) of KNN classifier.

Data setNDCRFSMIMIG-RFEIWFSCMIMDWFSCIFE

Lymphography38.334.7835.5935.5934.8835.2834.78
Dermatology97.76992.16492.16488.51290.7996.6887.139
Cardiotocography98.58998.40198.40198.40198.40198.58998.401
Pendigits97.91997.14597.14597.23897.50598.15997.625
Lung88.63688.06483.71276.39181.67887.68174.922
Carcinom85.4868.03732.25560.03565.8467.02631.952
Nci976.6975.4474.01269.02476.11948.42957.25
PCMAC87.64885.53886.15582.34884.76585.74378.952
Pixraw10P93.088.091.088.092.088.092.0
SMK-CAN-18770.01468.39369.00470.065.74768.42158.876
Lymphoma95.66784.72284.7569.80690.08372.05682.833
COIL2084.66280.73379.74371.66777.11472.02460.652
Average accuracy rate88.73484.2476.99475.58483.6476.50771.28
Wins/Ties/Losses12/0/012/0/012/0/012/0/012/0/012/0/0

The “Average” column gives the average accuracy value of the feature selection algorithm over all datasets. Bold represents the highest average classification prediction under this dataset.