scispace - formally typeset
Search or ask a question
Journal ArticleDOI

A new EDAS-based in-sample-out-of-sample classifier for risk-class prediction

11 Feb 2019-Management Decision (Emerald Publishing Limited)-Vol. 57, Iss: 2, pp 314-323
TL;DR: The proposed new non-parametric classifier delivers an outstanding predictive performance for a bankruptcy prediction application and is a real contender in actual applications in areas such as finance and investment, internet security, fraud and medical diagnosis.
Abstract: Nowadays, the field of data analytics is witnessing an unprecedented interest from a variety of stakeholders. The purpose of this paper is to contribute to the subfield of predictive analytics by proposing a new non-parametric classifier.,The proposed new non-parametric classifier performs both in-sample and out-of-sample predictions, where in-sample predictions are devised with a new Evaluation Based on Distance from Average Solution (EDAS)-based classifier, and out-of-sample predictions are devised with a CBR-based classifier trained on the class predictions provided by the proposed EDAS-based classifier.,The performance of the proposed new non-parametric classification framework is tested on a data set of UK firms in predicting bankruptcy. Numerical results demonstrate an outstanding predictive performance, which is robust to the implementation decisions’ choices.,The exceptional predictive performance of the proposed new non-parametric classifier makes it a real contender in actual applications in areas such as finance and investment, internet security, fraud and medical diagnosis, where the accuracy of the risk-class predictions has serious consequences for the relevant stakeholders.,Over and above the design elements of the new integrated in-sample-out-of-sample classification framework and its non-parametric nature, it delivers an outstanding predictive performance for a bankruptcy prediction application.

Summary (2 min read)

1. Introduction

  • Nowadays, the use of analytical methods in extracting intelligence from data is increasingly gaining popularity amongst a variety of public and private sectors’ stakeholders.
  • Predictive analytics techniques however are very popular in the financial industry, amongst many others, where high accuracy in risk-class predictions is crucial for decision-making, proactive planning, and prevention of potentially high losses.
  • Predictive Analytics techniques for class-belonging predictions fall into two main categories; namely, parametric methods and non-parametric methods, where non-parametric prediction methods have obvious advantages over parametric ones.
  • Evaluation Based on Distance from Average Solution (EDAS), first proposed by Keshavarz et al. (2015), is a multi-criteria method designed for ranking alternatives under multiple criteria.
  • The remainder of this paper unfolds as follows.

2. A New Integrated In-sample, Out-of-sample Classification Framework

  • The authors shall describe their integrated EDAS-based classification framework – see Figure 1 for a graphical representation of the process.
  • An observed riskclass membership, say 𝑌, is also available for all entities.
  • In sum, EDAS scores assign equal importance to positive and negative deviations from the average performer.
  • The choice of a decision rule for classification depends on the nature of the classification problem; that is, a two-class problem or a multi-class problem.

Initialization Step

  • 𝐸} in ascending order of distances and use the first 𝑘 entries in the list 𝐿𝑖1(1: 𝑘, . ) to classify 𝑒𝑛𝑡𝑖𝑡𝑦𝑖1 according to the chosen criterion; that is, the majority vote – see Table 2; in sum, 𝑒𝑛𝑡𝑖𝑡𝑦𝑖1 is assigned the predicted risk class label that the majority of its 𝑘 nearest neighbors have; } Output:.
  • This is the main reason for choosing a CBR framework for the out-of-sample classification instead of EDAS.

3. Empirical Results

  • In order to assess the performance of the proposed framework, the authors considered a sample of 6605 firm-year observations consisting of non-bankrupt and bankrupt UK firms listed on the London Stock Exchange (LSE) during 2010-2014 excluding financial firms and utilities as well as those firms with less than 5 months lag between the reporting date and the fiscal year.
  • As to the selection of the training sample and the test sample, the authors have chosen the size of the training sample to be twice the size of the test sample.
  • These drivers are current liabilities to total assets, No-credit interval – as measured by the number of days the company can continue to trade if it can no longer generate revenues, profit before tax to current liabilities, and current assets to total liabilities.
  • Since both the EDAS classifier and the k-NN classifier, trained on the in-sample classification obtained with EDAS, require a number of decisions to be made for their implementation, the authors considered several combinations of decisions to find out about the extent to which the performance of the proposed framework is sensitive or robust to these decisions.
  • The authors choices for these decisions are summarised in Table 2.

EDAS

  • Hereafter, the authors shall provide a summary of their empirical results and findings.
  • Table 3 provides a summary of In-sample statistics on the performance of the MDA model of Taffler (1984) reworked within the EDAS-CBR framework, which is an integrated in-sample-out-of-sample framework for EDAS-based classifiers.
  • These results show that the In-sample performance of the classifier is outstanding.
  • In fact, none of the non-bankrupt and bankrupt firms is misclassified.
  • As to bankrupt firms, on average, 99.60% to 99.99% are properly classified as shown by Sensitivity, depending on the choice of the distance metric to use within CBR, where the Mahalanobis distance seems to be the less desirable choice – although the difference in performance is marginal to recommend that the Mahalanobis distance should be avoided in implementing CBR.

4. Conclusions

  • The analytics toolbox of risk management is crucial for the financial industry amongst others.
  • The authors extended such toolbox with a new non-parametric classifier for predicting risk class belonging.
  • In addition, the basic concepts behind both EDAS and CBR are easy to explain to managers.
  • Last, but not least, the proposed classification framework delivers a very high performance similar to the DEA-based classifier proposed by Ouenniche and Tone (2017) and the MCDM classifiers proposed by Ouenniche et al. (2018a,b).
  • The difference between them however lies in how alternatives are rewarded or penalised for being close to or far from such reference point(s).

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

Edinburgh Research Explorer
A new EDAS-based in-sample-out-of-sample classifier for risk-
class prediction
Citation for published version:
Ouenniche, J, Uvalle Perez, OJ & Ettouhami, A 2018, 'A new EDAS-based in-sample-out-of-sample
classifier for risk-class prediction', Management Decision. https://doi.org/10.1108/MD-04-2018-0397
Digital Object Identifier (DOI):
10.1108/MD-04-2018-0397
Link:
Link to publication record in Edinburgh Research Explorer
Document Version:
Peer reviewed version
Published In:
Management Decision
General rights
Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s)
and / or other copyright owners and it is a condition of accessing these publications that users recognise and
abide by the legal requirements associated with these rights.
Take down policy
The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer
content complies with UK legislation. If you believe that the public display of this file breaches copyright please
contact openaccess@ed.ac.uk providing details, and we will remove access to the work immediately and
investigate your claim.
Download date: 09. Aug. 2022

A New EDAS-based In-sample-Out-of-sample Classifier for Risk-Class Prediction
Jamal Ouenniche
1
(Jamal.Ouenniche@ed.ac.uk)
University of Edinburgh, Business School
29 Buccleuch Place, Edinburgh EH8 9JS, United Kingdom
Oscar Uvalle (Oscar.Uvalle@ed.ac.uk)
University of Edinburgh, Business School
29 Buccleuch Place, Edinburgh EH8 9JS, United Kingdom
Aziz Ettouhami (touhami@fsr.ac.ma)
Conception and Systems Laboratory, Mohammed V University, Rabat, Morocco
Abstract
Purpose Nowadays, the field of data analytics is witnessing an unprecedented interest from a
variety of stakeholders. The purpose of this paper is to contribute to the subfield of predictive
analytics by proposing a new non-parametric classifier.
Design/methodology/approach The proposed new non-parametric classifier performs both in-
sample and out-of-sample predictions, where in-sample predictions are devised with a new EDAS-
based classifier, and out-of-sample predictions are devised with a CBR-based classifier trained on
the class predictions provided by the proposed EDAS-based classifier.
Findings The performance of the proposed new non-parametric classification framework is
tested on a dataset of UK firms in predicting bankruptcy. Numerical results demonstrate an
outstanding predictive performance, which is robust to the implementation decisions’ choices.
Practical implications The exceptional predictive performance of the proposed new non-
parametric classifier makes it a real contender in actual applications in areas such as finance and
investment, internet security, fraud, and medical diagnosis, where the accuracy of the risk-class
predictions has serious consequences for the relevant stakeholders.
1
Corresponding author: Jamal Ouenniche
Email address: Jamal.Ouenniche@ed.ac.uk

2
Originality/value Over and above the design elements of the new integrated in-sample-out-of-
sample classification framework and its non-parametric nature, it delivers an outstanding
predictive performance for a bankruptcy prediction application.
Keywords: In-sample Prediction, Out-of-sample Prediction, EDAS Classifier, CBR, k-Nearest
Neighbour Classifier, Bankruptcy, Risk Class Prediction
1. Introduction
Nowadays, the use of analytical methods in extracting intelligence from data is increasingly
gaining popularity amongst a variety of public and private sectors’ stakeholders. The popularity of
descriptive analytics techniques, predictive analytics techniques, and prescriptive analytics
techniques varies substantially from one industry to another. Predictive analytics techniques
however are very popular in the financial industry, amongst many others, where high accuracy in
risk-class predictions is crucial for decision-making, proactive planning, and prevention of
potentially high losses.
Predictive Analytics techniques for class-belonging predictions fall into two main categories;
namely, parametric methods and non-parametric methods, where non-parametric prediction
methods have obvious advantages over parametric ones. In this paper, we extend the toolbox of
non-parametric predictive methods by proposing a new integrated classifier that performs both in-
sample and out-of-sample predictions, where in-sample predictions are devised with a new EDAS-
based classifier, and out-of-sample predictions are devised with a Case-based Reasoning (CBR)-
based classifier trained on the class-belonging predictions provided by the proposed EDAS-based
classifier see Figure 1 for a snapshot of the design of the proposed prediction framework.
Evaluation Based on Distance from Average Solution (EDAS), first proposed by Keshavarz et
al. (2015), is a multi-criteria method designed for ranking alternatives under multiple criteria. This
MCDM method benchmarks all alternatives against a reference point; namely, the average
performer, and makes use of the positive and negative percentage deviations from the average
performer to construct an index for each alternative or entity , say
, which is then used to rank
alternatives. EDAS has been used to address a variety of applications; for example, building
construction (Turskis and Juodagalvienė, 2016); healthy and safe built environment (Zavadskas et
al., 2017); cultural heritage structures for renovation projects (Turskis et al., 2017b); conveyor
selection problem (Mathew and Sahu, 2018); automated guided vehicles selection problem

3
(Mathew and Sahu, 2018); steel rope analysis and diagnostic (Čereška el al., 2018); public
infrastructure for electric vehicles (Palevičius et al., 2018). The fuzzy version of EDAS
(Ghorabaee et al., 2016b) has also been used in many applications such as supplier selection
(Ghorabaee et al., 2016b; Keshavarz et al., 2017; Ghorabaee et al., 2017b); subcontractor
evaluation (Ghorabaee et al., 2017a; Keshavarz et al., 2018); facility location (Ghorabaee et al.,
2016a); selection of solid waste disposal sites (Kahraman et al., 2017); construction equipment
evaluation (Ghorabaee et al., 2018); personnel selection (Turskis et al., 2017a). Although
Keshavarz et al. (2015) applied EDAS to an inventory classification problem, no details on the
classification rule(s) have been provided. In this paper, we propose a new in-sample classifier
based on EDAS and hybridise it with CBR as an out-of-sample classifier, which is trained on the
class-belonging predictions provided by EDAS.
Figure 1: An Integrated EDAS-CBR Framework
for In-Sample and Out-of-Sample Class-belonging Predictions
The remainder of this paper unfolds as follows. In section 2, we provide a detailed description
of the proposed integrated in-sample and out-of-sample framework for EDAS-based classifiers

4
and discuss implementation decisions. In section 3, we empirically test the performance of the
proposed framework in bankruptcy prediction of companies listed on the London Stock Exchange
(LSE) and report on our findings. Finally, section 4 concludes the paper.
2. A New Integrated In-sample, Out-of-sample Classification Framework
In this section, we shall describe our integrated EDAS-based classification framework see Figure
1 for a graphical representation of the process. Without loss of generality, we shall customize the
presentation of the proposed framework to a bankruptcy application as follows:
Input: A set of alternatives or entities (e.g., LSE listed firm-year observations) to be assessed
on pre-specified criteria (e.g., financial criteria) along with their measures (e.g., financial
ratios), where the measure of each criterion could either be minimized or maximized. Thus,
each entity, say (), is represented by an -dimensional vector of (observed)
measures of the criteria under consideration, say
󰇛

󰇜, where

denote the observed
measure of criterion for entity , and the set of
s shall be denoted by . An observed risk-
class membership, say , is also available for all entities. The historical sample is divided
into a training sample, say
, and a test sample, say
, where 
(resp. 
) denote the
cardinality of
(resp.
).
Phase 1: EDAS-based In-Sample Classifier
Step 1: Computation of a Reference Point or Benchmark
The reference point is a virtual benchmark representing the average performance across all
alternatives in the training sample
(
) on each criterion (). Such
average performer or benchmark is represented by a virtual alternative described by an -
dimensional vector, say , where entry corresponds to the average performance on criterion
and computed as follows:



; 
where

denote the observed performance of alternative
on criterion ().
Step 2: Computation of Individual Positive and Negative Percentage Deviations from The
Reference Point or Benchmark with respect to Each Criterion
Compute the positive and negative distances with respect to each criterion , say
󰇛

󰇜
and
󰇛

󰇜
, between each alternative in the training sample
(
) and the virtual

Citations
More filters
Posted Content
01 Nov 2017-viXra
TL;DR: The proposed intuitionistic fuzzy EDAS method, which is based on the data belonging to membership, nonmembership, and hesitance degrees, is applied to the evaluation of solid waste disposal site selection alternatives.
Abstract: Evaluation based on Distance from Average Solution (EDAS) is a new multicriteria decision making (MCDM) method, which is based on the distances of alternatives from the average scores of attributes.

120 citations

Posted Content
01 Mar 2018
TL;DR: A new integrated model is proposed for supplier evaluation and order allocation which considers both environmental and economic factors and shows that the proposed model is efficient and applicable for real-world problems.
Abstract: A new approach is proposed for supplier evaluation and order allocation.Environmental and economic criteria are considered in the proposed approach.Interval type-2 fuzzy sets and the EDAS method are used for supplier evaluation.A multi-objective linear programming is proposed for order allocation.A sensitivity analysis is made to examine effects of environmental criteria on the model. Nowadays environmental performance of suppliers becomes more important because of competitive conditions. Besides, the economic performance has been a significant factor for companies to choose their suppliers. In this paper, a new integrated model is proposed for supplier evaluation and order allocation which considers both environmental and economic factors. We use the EDAS (Evaluation based on Distance from Average Solution) method and interval type-2 fuzzy sets for evaluation of suppliers with respect to environmental criteria. According to this evaluation two parameters are defined for each supplier: positive score and negative score. These parameters, together with cost parameters, are utilized to propose a multi-objective mathematical model for determination of order quantity from each supplier. A numerical example is used in this paper to show the applicability of the proposed integrated model. Also, a sensitivity analysis is made to examine the effect of weighting environmental criteria on total purchasing cost and quantity of order from each supplier. The results show that the proposed model is efficient and applicable for real-world problems.

114 citations

Journal ArticleDOI
TL;DR: A three-stage methodology has been applied for presenting a framework for logistics center location selection in the context of Kayseri's logistics development plan and the proposed methodology BWM–EDAS and also EDAS with different distance measures, which are applied for the first time in the literature, provides helpful findings to rank the logistics center locations.
Abstract: A logistics center is the hub of a specific area, within various logistics-related activities (distribution, storage, transportation, consolidation, handling, customs clearance, imports, exports, transit processes, infrastructural services, insurance, banking, etc.) that are performed on a commercial basis. Determining the location of the logistics center is an important decision regarding cost and benefit analysis. A three-stage methodology has been applied for presenting a framework for logistics center location selection in the context of Kayseri’s logistics development plan. The first stage includes the determination of criteria through literature review and interviews with experts. The second stage includes the weighting of determined criteria using linear BWM (best–worst method). The third stage includes the ranking of locations using the evaluation based on distance from average solution (EDAS) method with different distance measures. Our proposed methodology BWM–EDAS and also EDAS with different distance measures, which are applied for the first time in the literature, provides helpful findings to rank the logistics center locations. Lastly, sensitivity analysis is conducted to validate the robustness.

38 citations

Journal ArticleDOI
TL;DR: The hybrid method of best-worst method, Multi-Objective Optimization by Ratio Analysis plus Full Multiplicative Form (MULTIMOORA), and Evaluation based on Distance from Average Solution (EDAS) is used to evaluate new energy vehicles (NEVs) and select the best new energy vehicle.
Abstract: With resource shortage and environmental pollution becoming more and more serious, the development of new energy vehicles (NEVs) plays an important role. In this paper, the hybrid method of best-worst method (BWM), Multi-Objective Optimization by Ratio Analysis plus Full Multiplicative Form (MULTIMOORA), and Evaluation based on Distance from Average Solution (EDAS) is used to evaluate new energy vehicles (NEVs) and select the best new energy vehicle. BWM method is used to obtain the subjective preference weight, MULTIMOORA method is used to integrate the objective data with the subjective weight to evaluate new energy vehicles, and the final ranking of alternatives is obtained by the EDAS method. The paper collect the data of 22 representative new energy vehicle types in China, the validity and feasibility of the method is verified.

14 citations

Journal ArticleDOI
TL;DR: The findings conclude that the proposed new classifier can deliver a very high predictive performance, which makes it a real contender in industry applications in finance and investment.
Abstract: Nowadays, business analytics has become a common buzzword in a range of industries, as companies are increasingly aware of the importance of high quality predictions to guide their pro-active planning exercises. The financial industry is amongst those industries where predictive analytics techniques are widely used to predict both continuous and discrete variables. Conceptually, the prediction of discrete variables comes down to addressing sorting problems, classification problems, or clustering problems. The focus of this paper is on classification problems as they are the most relevant in risk-class prediction in the financial industry. The contribution of this paper lies in proposing a new classifier that performs both in-sample and out-of-sample predictions, where in-sample predictions are devised with a new VIKOR-based classifier and out-of-sample predictions are devised with a CBR-based classifier trained on the risk class predictions provided by the proposed VIKOR-based classifier. The performance of this new non-parametric classification framework is tested on a dataset of firms in predicting bankruptcy. Our findings conclude that the proposed new classifier can deliver a very high predictive performance, which makes it a real contender in industry applications in finance and investment.

11 citations

References
More filters
Journal ArticleDOI
TL;DR: A new method of Evaluation based on Distance from Average Solution (EDAS) is introduced for multi-criteria inventory clas- sification (MCIC) problems and shows that the proposed method is stable in different weights and well consistent with the other methods.
Abstract: An effective way for managing and controlling a large number o f inventory items or stock keeping units (SKUs) is the inventory classification. Tradi tional ABC analysis which based on only a single criterion is commonly used for classification of SKU s. However, we should consider inven- tory classification as a multi-criteria problem in practice . In this study, a new method of Evaluation based on Distance from Average Solution (EDAS) is introduced for multi-criteria inventory clas- sification (MCIC) problems. In the proposed method, we use po sitive and negative distances from the average solution for appraising alternatives (SKUs). To represent performance of the proposed method in MCIC problems, we use a common example with 47 SKUs. Comparing the results of the proposed method with some existing methods shows the good performance of it in ABC classifica- tion. The proposed method can also be used for multi-criteria decision-making (MCDM) problems. A comparative analysis is also made for showing the validity and stability of the proposed method in MCDM problems. We compare the proposed method with VIKOR, TOPSIS, SAW and COPRAS methods using an example. Seven sets of criteria weights and Spearman's correlation coefficient are used for this analysis. The results show that the proposed method is stable in different weights and well consistent with the other methods.

696 citations

Journal ArticleDOI
TL;DR: The results of this study show that the extended fuzzy EDAS method is efficient and has good tability for solving MCDM problems.
Abstract: In the real-world problems, we are likely confronted with some alternatives that eed to be evaluated with respect to multiple conflicting criteria. Multi-criteria ecision-making (MCDM) refers to making decisions in such a situation. There are any methods and techniques available for solving MCDM problems. The evaluation ased on distance from average solution (EDAS) method is an efficient multi-criteria ecision-making method. Because the uncertainty is usually an inevitable part of he MCDM problems, fuzzy MCDM methods can be very useful for dealing with the eal-world decision-making problems. In this study, we extend the EDAS method o handle the MCDM problems in the fuzzy environment. A case study of supplier election is used to show the procedure of the proposed method and applicability of t. Also, we perform a sensitivity analysis by using simulated weights for criteria to xamine the stability and validity of the results of the proposed method. The results f this study show that the extended fuzzy EDAS method is efficient and has good tability for solving MCDM problems.

243 citations

Journal ArticleDOI
TL;DR: In this article, a critical appraisal of existing UK Z-score models and their operational utility is presented, and two previously unpublished models are described which address respectively the need for separate models for manufacturing and distribution companies, and the utility of the jackknife discriminant approach in practice.
Abstract: This paper has four main aims. Firstly it attempts a critical appraisal of extant UK Z-score models and seeks to assess in each case their operational utility. Next two previously unpublished models are described which address respectively (i) the need for separate models for manufacturing and distribution companies, and (ii) the utility of the jackknife discriminant approach in practice. Then developments of the technique to enhance considerably its utility to the practitioner are described. Finally a brief review of how such approaches are currently being used in the UK and by whom is provided.

235 citations

Journal ArticleDOI
TL;DR: Evaluation based on Distance from Average Solution (EDAS) is a new multicriteria decision making (MCDM) method, which is based on the distances of alternatives from the average scores of attributes as discussed by the authors.
Abstract: Evaluation based on Distance from Average Solution (EDAS) is a new multicriteria decision making (MCDM) method, which is based on the distances of alternatives from the average scores of attributes...

195 citations

Journal ArticleDOI
TL;DR: In this article, a hybrid multi-criteria decision-making (MCDM) approach is proposed to deal with this evaluation process in the fuzzy environment, where the subjective and objective criteria weights are combined to determine more justified weights for criteria.

139 citations

Frequently Asked Questions (1)
Q1. What are the contributions in this paper?

The purpose of this paper is to contribute to the subfield of predictive analytics by proposing a new non-parametric classifier. The proposed new non-parametric classifier performs both insample and out-of-sample predictions, where in-sample predictions are devised with a new EDASbased classifier, and out-of-sample predictions are devised with a CBR-based classifier trained on the class predictions provided by the proposed EDAS-based classifier.