Data Cleansing Meets Feature Selection: A Supervised Machine Learning Approach

摘要

This paper presents a novel procedure to apply in a sequentialway two data preparation techniques from a different nature such asdata cleansing and feature selection. For the former we have experiencedwith a partial removal of outliers via inter-quartile range whereas forthe latter we have chosen relevant attributes with two widespread featuresubset selectors like CFS (Correlation-based Feature Selection) andCNS (Consistency-based Feature Selection), which are founded on correlationand consistency measures, respectively. Empirical results on sevendifficult binary and multi-class data sets, that is, with a test error rate ofat least a 10%, according to accuracy, with C4.5 or 1-nearest neighbourclassifiers without any kind of prior data pre-processing are outlined.Non-parametric statistical tests assert that the meeting of the aforementionedtwo data preparation strategies using a correlation measure forfeature selection with C4.5 algorithm is significant better, measured withroc measure, than the single application of the data cleansing approach.Last but not least, a weak and not very powerful learner like PARTachieved promising results with the new proposal based on a consistencymeasure and is able to compete with the best configuration of C4.5. Tosum up, bearing in mind the new approach, for roc measure PART classifierwith a consistency metric behaves slightly better than C4.5 and acorrelation measure