Identification of interaction patterns and classification with applications to microarray data

Abstract

Emerging patterns represent a class of interaction structures which has been recently proposed as a tool in data mining. In this paper, a new and more general definition refering to underlying probabilities is proposed. The defined interaction patterns carry information about the relevance of combinations of variables for distinguishing between classes. Since they are formally quite similar to the leaves of a classification tree, we propose a fast and simple method which is based on the CART algorithm to find the corresponding empirical patterns in data sets. In simulations, it can be shown that the method is quite effective in identifying patterns. In addition, the detected patterns can be used to define new variables for classification. Thus, we propose a simple scheme to use the patterns to improve the performance of 1 classification procedures. The method may also be seen as a scheme to improve the performance of CARTs concerning the identification of interaction patterns as well as the accuracy of prediction.2

Statistics for proteomics: Experimental design and 2-DE differential analysis

“…Further investigations using the results issued from such a collaboration can be considered with increased confidence. Statistical tools such as discriminate analysis, regression methods or supervised classification [50][51][52][53][54][55] can be further applied to accurately discriminate the status of unknown samples, normal or pathologic for instance. The interaction schema between statisticians and biologists is particularly important for the detection of differentially expressed proteins involved in pathologies since it can lead to the discovery of biomarker candidates.…”

Gaining New Insight into Machine-Learning Datasets via Multiple Binary-Feature Frequency Ranks with a Mobile Benign/Malware Apps Example

“…Typical usage of the Kruskal-Wallis test in machine-learning is using as a filtering method for feature selection in high-dimensional datasets [27,28]. It is appropriate for not only binary classification but also multi-class classification problems [29]. The literature has successfully used the test on analyzing and comparing data with different characteristics, for example, censored data [30] and microarray gene expression data [31], but also addressed the limitations when applied in high dimensional low sample size data (shallow datasets) [32].…”

WilcoxCV: an R package for fast variable selection in cross-validation

“…In binary classification, it is usual to rank genes according to the P-value obtained in, e.g. the t-test for two independent samples and related methods or the Wilcoxon rank sum test, also known as the Mann-Whitney test (Boulesteix and Tutz, 2006;Dettling and Bu¨hlmann, 2003). The genes with the smallest P-values are then selected and used for classifier construction.…”