Partition-conditional ICA for Bayesian classification of microarray data

Accurate classification of microarray data is very important for medical decision making. Past studies have shown that class-conditional independent component analysis (CC-ICA) is capable of improving the performance of naïve Bayes classifier in microarray data analysis. However, when a microarray dataset has a small number of samples for some classes, the application of CC-ICA may become infeasible. This paper extends CC-ICA and proposes a partition-conditional independent component analysis (PC-ICA) method for naive Bayes classification of microarray data. Compared to ICA and CC-ICA, PC-ICA represents an in-between concept for feature extraction. Our experimental results on two microarray datasets show that PC-ICA is more effective than ICA in improving the performance of naïve Bayes classification of microarray data.