Showing papers on "Rand index published in 2003"

Efficient Clustering Methods for Tumor Classification With Microarrays

Abstract: Microarray experiments yield expression measurements of thousands of genes simultaneously. One important application is the classification of cancer tumor types. For this task, we present a comparison of hierarchical clustering algorithms, PAM, fanny, k-means and SOM’s on two real microarray data sets. Several distance measures are used to build dissimilarity matrices from the data. We introduce the new resampling approach bootstrap-scaling (BOSe) to simulate different amounts of noise in the data. Based on the simulation results, the k-means algorithm and the Manhattan distance give the best results.