A novel hybrid undersampling method for mining unbalanced datasets in banking and insurance

Sundarkumar, G. Ganesh; Ravi, Vadlamani

doi:10.1016/j.engappai.2014.09.019

Cited by 103 publications

(33 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Apart from the main implementations on transaction data, the authors in [28] focused on phishing detection from official banking websites and applied a multi-label classifier based associative classification DM for effective detection of phishing in websites with high levels of accuracy. In order to improve the customer credit card churning prediction for a Latin-American bank, the authors in [29] adopted improved DM techniques that are based on K-means clustering and support vector machines (SVM). Blog mining (text mining and cluster analysis) was applied in [30], where security risks, protection strategy and security trends of mobile banking were summarized from more than 200,000 results of the Google blog search engine.…”

Section: Security and Fraud Detectionmentioning

confidence: 99%

Digitalisation and Big Data Mining in Banking

Hassani

Huang

Silva

2018

BDCC

115

View full text Add to dashboard Cite

Banking as a data intensive subject has been progressing continuously under the promoting influences of the era of big data. Exploring the advanced big data analytic tools like Data Mining (DM) techniques is key for the banking sector, which aims to reveal valuable information from the overwhelming volume of data and achieve better strategic management and customer satisfaction. In order to provide sound direction for the future research and development, a comprehensive and most up to date review of the current research status of DM in banking will be extremely beneficial.Since existing reviews only cover the applications until 2013, this paper aims to fill this research gap and presents the significant progressions and most recent DM implementations in banking post 2013. By collecting and analyzing the trends of research focus, data resources, technological aids, and data analytical tools, this paper contributes to bringing valuable insights with regard to the future developments of both DM and the banking sector along with a comprehensive one stop reference table. Moreover, we identify the key obstacles and present a summary for all interested parties that are facing the challenges of big data.

show abstract

Section: Security and Fraud Detectionmentioning

confidence: 99%

Digitalisation and Big Data Mining in Banking

Hassani

Huang

Silva

2018

BDCC

115

View full text Add to dashboard Cite

show abstract

“…Some of these are: size, noise, sparsity and uncertainty. Furthermore, in the vast majority of financial applications, data is highly unbalanced [24]. For example, in credit card applications the number of good customers is much higher than that of bad customers, and in fraud detection the majority of the data is normal transactions with only a few fraudulent transactions.…”

Section: [1]-[3])mentioning

confidence: 99%

Multiobjective Evolutionary Optimization of Type-2 Fuzzy Rule-Based Systems for Financial Data Classification

Antonelli

Bernardo

Hagras

et al. 2017

IEEE Trans. Fuzzy Syst.

View full text Add to dashboard Cite

Abstract-Classification techniques are becoming essential in the financial world for reducing risks and possible disasters. Managers are interested in not only high accuracy but also in interpretability and transparency. It is widely accepted now that the comprehension of how inputs and output are related to each other is crucial for taking operative and strategic decisions. Furthermore, inputs are often affected by contextual factors and characterized by a high level of uncertainty. In addition, financial data are usually highly skewed towards the majority class. With the aim of achieving high accuracies, preserving the interpretability and managing uncertain and unbalanced data, the paper presents a novel method to deal with financial data classification by adopting type-2 fuzzy rule-based classifiers (FRBCs) generated from data by a multi-objective evolutionary algorithm (MOEA). The classifiers employ an approach, denoted as scaled dominance, for defining rule weights in such a way to help minority classes to be correctly classified. In particular, we have extended PAES-RCS, an MOEA-based approach to learn concurrently the rule and data bases of FRBCs, for managing both interval type-2 fuzzy sets and unbalanced datasets. To the best of our knowledge, this is the first work that generates type-2 FRBCs by concurrently maximizing accuracy and minimizing the number of rules and the rule length with the objective of producing interpretable models of real-world skewed and incomplete financial datasets. The rule bases are generated by exploiting a rule and condition selection (RCS) approach, which selects a reduced number of rules from a heuristically generated rule base and a reduced number of conditions for each selected rule during the evolutionary process. The weight associated with each rule is scaled by the scaled dominance approach on the fuzzy frequency of the output class, in order to give a higher weight to the minority class. As regards the data base learning, the membership function parameters of the interval type-2 fuzzy sets used in the rules are learned concurrently to the application of RCS. Unbalanced datasets are managed by using, in addition to complexity, selectivity and specificity as objectives of the MOEA rather than only the classification rate. We tested our approach, named IT2-PAES-RCS, on eleven financial datasets and compared our results with the ones obtained by the original PAES-RCS with three objectives and with and without scaled dominance, the fuzzy rule-based classifiers FARC-HD and FURIA, the classical C4.5 decision tree algorithm and its costsensitive version. Using non-parametric statistical tests, we will show that IT2-PAES-RCS generates FRBCs with, on average, accuracy statistically comparable to and complexity lower than the ones generated by the two versions of the original PAES-RCS. Further, the FRBCs generated by FARC-HD and FURIA and the decision trees computed by C4.5 and its cost-sensitive version, despite the highest complexity, result to be less accurate than the FRBCs...

show abstract

“…Correlation analysis is given a set of Item and a record collection, through the analysis of the recorded set, to find the correlation between the items. As an important part of the correlation analysis, the support degree and the reliability are the important contents of the correlation analysis [11][12][13]. If the association rules satisfy the minimum support threshold and the minimum confidence threshold, then it is considered that the association rule is meaningful.…”

Section: Correlation Analysismentioning

confidence: 99%