Comparison of PCA with ICA from data distribution perspective
This is an incremental study providing empirical insights for researchers in signal processing or data analysis on algorithm performance under different data conditions.
The authors compared ICA and PCA on simulated noisy time series with varying distribution parameters and noise levels, finding that ICA generally performs better by considering higher moments of data distribution, while PCA remains sensitive to signal correlations.
We performed an empirical comparison of ICA and PCA algorithms by applying them on two simulated noisy time series with varying distribution parameters and level of noise. In general, ICA shows better results than PCA because it takes into account higher moments of data distribution. On the other hand, PCA remains quite sensitive to the level of correlations among signals.