Internal Evaluation Measures as Proxies for External Indices in Clustering Gene Expression Data

Abstract

Abstract-Several external indices that use information not present in the dataset were shown to be useful for evaluation of representative based clustering algorithms. However, such supervised measures are not directly useful for construction of better clustering algorithms when class labels are not provided. We propose a method for identifying internal cluster evaluation measures that use only information present in the dataset and are related to given external indices. We utilize these internal measures for the construction of representative based clustering algorithms. Both identification and utilization steps of the proposed method are enabled by use of a component-based clustering algorithm design. Experiments on 432 algorithms using gene expression data sets provide evidence that some internal measures could be used as surrogates for external indices proposed in the literature. Moreover, the obtained results suggest that internal measures correlated to selected external indices can guide the algorithms toward significantly better cluster models.

A method for design of data-tailored partitioning algorithms for optimizing the number of clusters in microarray analysis

“…However, structure of the best RC based algorithms (Table 5) suggests that on different datasets, different RCs participated in building the best algorithm. From Table 5, it can be noticed that AIC and SILHOU are the only RCs used for "Evaluate clusters" sub-problems (which is a result also suggested by [7], ONLINE was the most frequent for "Update representatives". Still, detailed inspection of the results showed that there were a lot of algorithms that showed minimal difference in performance from the best one, but had a quite different structure.…”

“…This research extends our research of RC-based partitioning clustering algorithms on microarray data analysis [7], where the main focus was identification of algorithm structures that can reveal the right clustering structure, given a known K. In comparison to [7] this research makes the following extensions:…”

“…In [7], partitioning (k-means like) clustering algorithms based on the idea of RC-based design were used to analyze algorithm structure that adapts best to dataset "ground truth". Those algorithms were unable to find an optimal K for the data at hand.…”

“…• Proposal of a method for identifying the most-suited divisive clustering algorithms, when K is not known in advance (algorithms investigated in [7] needed user defined number of clusters).…”

Consensus Clustering for Microarray Gene Expression Data

“…To evaluate the performance of our clustering algorithm for real datasets we have considered the datasets used by Monti et al (2003). These datasets are often used as benchmarks to evaluate the cluster analysis techniques for microarray data (Vukicevic et al 2011), Table 1 shows the real datasets used and their respective number of clusters. V.…”