Frits Dannenberg

Sparse principal components analysis

 

Frits Dannenberg (2010-09-08)

 

Principal component analysis (PCA) is a widespread exploratory data analysis tool. Sparse principal component analysis (SPCA) is a method that improves upon PCA by increasing the number of zeros in the loading vectors of PCA results. This makes the results more understandable and more usable. This bachelor's thesis introduces both methods, and includes examples using both real-world data and artifcial data. Also, the behavior of PCA under departure from weakly stationary data is explored.