Threshold Tuning for Improved Classification Association Rule Mining

Abstract

Abstract. One application of Association Rule Mining (ARM) is to identify Classification Association Rules (CARs) that can be used to classify future instances from the same population as the data being mined. Most CARM methods first mine the data for candidate rules, then prune these using coverage analysis of the training data. In this paper we describe a CARM algorithm that avoids the need for coverage analysis, and a technique for tuning its threshold parameters to obtain more accurate classification. We present results to show this approach can achieve better accuracy than comparable alternatives at lower cost.

Multi-Party Argument from Experience

“…PISA has been fully implemented in Java, and uses the tool described in [5] to mine its association rules (ARs). We have applied PISA to several datasets including the welfare benefits application used to evaluate PISA [18] and a datsset drawn from an application to process applications for a nursery school in Ljubljana [12].…”

A Hybrid Statistical Data Pre-processing Approach for Language-Independent Text Classification

“…associative classification [2]. Coenen et al [5] and Shidara et al [24] indicate that results presented in the studies of [15,16,28] show that in many cases associative classification offers greater classification accuracy than other classification approaches, such as C4.5 [19] and RIPPER (Repeated Incremental Pruning to Produce Error Reduction) [7].…”

“…The evaluation of Hybrid DIAAF/GSSC, under a statistical "bag of phrases" DR setting, is conducted using the TFPC (Total From Partial Classification) [5] associative classification algorithm; although any other associative classifier generator could equally well have been used. The experimental results demonstrate that Hybrid DIAAF/GSSC based textual data pre-processing approach outperforms alternative techniques with respect to the accuracy of classification.…”

Statistical Identification of Key Phrases for Text Classification

“…have demonstrated that, for many classification problems, CARM approaches can lead to better classification accuracy than other methods. Earlier work by the authors [2] [3], employing a CARM algorithm, TFPC, showed that appropriate selection of thresholds led to high classification accuracy in a wide range of cases. In the present work we seek to apply this algorithm to the TC problem, and to identify parameter values to optimise its accuracy.…”

“…All experiments described in this paper were undertaken using the authors' TFPC algorithm [2] [3]. TFPC (Total From Partial Classification) is a CARM algorithm that constructs a classifier by identifying Classification Association Rules (CARs) from a set of previously-classified cases.…”