Customer classification: A Mamdani fuzzy inference system standpoint for modifying the failure mode and effect analysis based three dimensional approach
Introduction
Of the paramount characteristics ubiquitous amongst almost all thriving enterprises is offering class specific value prepositions to each category of customers. Such a strategy is usually initiated with customer classification, a concept evolving to be integral to marketing and being applied by numerous firms to meet customer needs more appropriately (Floh, Zauner, Koller, & Rusch, 2014). Customer classification is also a prerequisite for targeting the best customer segment (Madzík & Shahin, 2020) and predicting the customer churn (Abbasimehr, Setak, & Soroor, 2013). Literature review indicates that a considerable number of such classification efforts have been made through the lens of the customer loyalty level.
Loyalty is a strong commitment to consistently re-purchase or re-patronize a preferred product or service in the future, which, as a result, leads to repetition of buying the same set of the brand, in spite of the fact that situational impacts and marketing attempts may potentially result in switching behavior (Oliver, 1999, p. 34). This definition elucidates the fact that loyalty serves as a shield protecting sellers against their rivals (Mascarenhas, Kesavan, & Bernacchi, 2006), and (truly) loyal customers are committed to keep re-purchasing regardless of circumstance (Narayandas, 2005). Moreover, such customers are greatly cooperative and willing to provide positive word of mouth and recommendation (Tanford & Baloglu, 2013).
While there is not a universal approach for measuring loyalty, literature shows several methods employed for the purpose, such as Net Promoter Score (Reichheld, 2003), Loyalty Ladder (Narayandas, 2005, Mascarenhas et al., 2006), Bandyopadhyay and Martell’s (2007) framework for consumer brand loyalty classification, and Loyalty Matrix used by Tanford and Baloglu (2013). In a recent study, Madzík & Shahin (2020) places a more explicit emphasis on including behavioral factors of loyalty as necessary supplements to the attitudinal one. In other words, inspired by the Risk Priority Number (RPN) calculation from multiplication of three risk factors in Failure Mode and Effect Analysis (FMEA) – a well-known technique in risk assessment and quality management – they proposed computing the Loyalty Priority Number (LPN) from the product of three loyalty dimensions: customer purchase value (V), purchase frequency (F) and loyalty (L). The LPN was, in turn, utilized as an overarching index for customer classification. However, the multiplication formula of basic FMEA, adopted in the Madzík & Shahin’s (2020) Three Dimensional Customer Classification (TDCC) methodology, has widely been criticized in literature. The TDCC limitations, together with their corresponding FMEA roots (Liu et al., 2012, Liu et al., 2013, Geramian et al., 2017, Geramian et al., 2019, Geramian et al., 2020), are summarized in Table 1.
Since these issues are of a FMEA nature, the present research utilizes the most popular approach for modifying the traditional FMEA shortcomings, this time, to resolve the TDCC limitations. This popular solution is the expert system and artificial intelligence approach of Fuzzy Inference System or FIS (Liu et al., 2013, Geramian et al., 2017, Geramian et al., 2019, Geramian et al., 2020). FIS is capable of resolving almost all of the abovementioned issues (Geramian et al., 2017, Geramian et al., 2019, Geramian et al., 2020). Moreover, as ratings of TDCC loyalty dimensions are determined through respondents’ judgment and prone to uncertainties hidden in human judgments, application of the fuzzy sets theory (Zadeh, 1965), in general, and FIS fuzzification (Geramian et al., 2017, Geramian et al., 2019, Geramian et al., 2020) and fuzzy numbers (Shahin, Barati, & Geramian, 2017), in particular, are of tremendous advantages in this regard. With respect to the fact that TDCC is of three inputs (loyalty dimensions) and one output (LPN), along with a questionable or unknown inputs to output relationship, we draw on a Multi Input Single Output (MISO) FIS of the Mamdani type (Mamdani and Assilian, 1975, Geramian et al., 2019, Geramian et al., 2020). Using a type 1 fuzzy system – versus more enhanced versions such as the type 2 fuzzy logic – the present study will act as an initial step in application of fuzzy inference in the TDCC problem. Therefore, it is mainly application oriented, by nature.
As a result, the objective of the present research is to utilize the Mamdani MISO FIS methodology to modify TDCC. For this purpose, first we identify four potential market types, as contextual scenarios, which customer classification analysts may encounter. Then, corresponding to them, four tailored FIS’s based on TDCC are developed. It is noteworthy that classes/categories of the proposed approach are designed on the basis of the Loyalty Matrix methodology used by Tanford and Baloglu (2013). Implementation of the developed approach in the identified scenarios facilitated highlighting outperformance of FIS based TDCC over basic TDCC from several aspects.
Academically, this study will equip the customer classification function of customer relationship management with the expert and intelligent system of FIS. To the best of the authors’ knowledge, the present study is the first research applying the fuzzy logic and, in particular, the Mamdani MISO FIS, to modify the FMEA based customer classification. In other words, such application will bring a higher level of accuracy to this specific manner of the classification problem through modifying the drawbacks of its underlying method – the traditional FMEA. As a result, integration of both the behavioral and attitudinal dimensions into a single classification oriented index will be carried out much more reliably.
Practically, it will shed light on the fact that the context matters when classifying customers. Identifying the potential contextual factors, this paper will draw a comprehensive picture of market scenarios that practitioners may face. Besides, it will provide four FIS’s, with each being tailor made to classify customers in a specific scenario. Accordingly, this research proposes a viable methodology that helps companies tailor their offerings and marketing strategies towards each specific customer class.
Section snippets
Loyalty based customer classification approaches
By measuring the extent to which a customer is willing to recommend a given company to others, Reichheld (2003) developed Net Promoter Score as a basis for customer categorization. The measure is computed via subtracting the percentage of detractors from that of promoters and used to classify customers into detractors, passively satisfied and promoters. Narayandas (2005) categorizes the customers’ loyalty behaviors in the business market setting into a hierarchy of levels called Loyalty Ladder.
Three dimensional customer classification
Madzík & Shahin (2020) developed TDCC, the general mechanism of which is based on the traditional FMEA. FMEA is an analytic approach for definition, identification and elimination of potential and/or known failures, errors and so forth from design, service, process and system, prior to the items reaching the customer (Stamatis, 1995, Liu et al., 2012). In traditional FMEA, risk of failures is prioritized through the RPN index, calculable via Eq. (1) (Qin, Xi, & Pedrycz, 2020).
Where
Proposed approach: FIS based three dimensional customer classification
Our proposed methodology has a framework (Fig. 1) inspired by the fuzzy FMEA methodology (Geramian et al., 2019, Geramian et al., 2020). In Fig. 1., as the direct relationship between the loyalty dimensions (V, F and L) and LPN is questionable in TDCC, it is depicted using a dotted arrow between crisp values of these inputs and the output value, which is indicated using the Defuzzified form of a Fuzzy LPN (Defuzzified FLPN).
Instead, a mapping between them can be established through FIS,
Application of the approach
In order to investigate applicability and effectiveness of the proposed methodology, it is applied in the context of some scenarios by considering several assumptive customers, each being characterized by three assumptive values for the loyalty dimensions. This provided us with the chance of implementing our proposed approach in comparison to TDCC, followed by findings and discussion.
Practical implications
This study sheds light on the fact that marketing managers should carry out the customer classification task with respect to the context – market type – they are planning for; otherwise, they may end up with misleading results. In competitive markets, managers should pay much more attention to attitudinal loyalty factors. In such contexts, a given company’s customers with only high behavioral loyalty – purchase value or frequency – indication may easily stop buying from the company because they
Conclusions
In the present research, we modified TDCC, an approach recently developed to classify customers, by means of the MISO Mamdani FIS and, in part, Loyalty Matrix. Benefits of such modification are extensive: tackling human judgment uncertainties extant within input data; providing both within and between scenario discriminability; considering relative importance weights of the loyalty dimensions; facilitating a relaxation of the questionable multiplication formula; introducing a gap free output
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors are really thankful to the anonymous reviewers whose invaluable comments facilitated substantial improvement in the quality of this research.
References (41)
- et al.
Customer churn prediction in telecommunication industry using data certainty
Journal of Business Research.
(2019) - et al.
Does attitudinal loyalty influence behavioral loyalty? A theoretical and empirical study
Journal of Retailing and Consumer Services
(2007) - et al.
A generalized type-2 fuzzy granular approach with applications to aerospace
Information Sciences.
(2016) - et al.
Hybrid intelligent system for cardiac arrhythmia classification with Fuzzy K-Nearest Neighbors and neural networks combined with a fuzzy system
Expert Systems with Applications.
(2012) - et al.
Customer segmentation using unobserved heterogeneity in the perceived-value–loyalty–intentions link
Journal of Business Research.
(2014) - et al.
An extended FMEA approach based on the Z-MOORA and fuzzy BWM for prioritization of failures
Applied Soft Computing.
(2019) - et al.
A rough set-based interval-valued intuitionistic fuzzy conceptual design decision approach with considering diverse customer preference distribution
Advanced Engineering Informatics.
(2021) - et al.
Diagnosis of feedwater heater performance degradation using fuzzy inference system
Expert Systems with Applications.
(2017) - et al.
Risk evaluation approaches in failure mode and effects analysis: A literature review
Expert Systems with Applications.
(2013) - et al.
Risk evaluation in failure mode and effects analysis with extended VIKOR method under fuzzy environment
Expert Systems with Applications.
(2012)
A generic two-stage fuzzy inference system for dynamic prioritization of customers
Expert Systems with Applications.
An experiment in linguistic synthesis with a fuzzy logic controller
International Journal of Man-Machine Studies.
Application of interval type-2 fuzzy logic systems to gas turbine fault diagnosis
Applied Soft Computing.
Failure mode and effects analysis (FMEA) for risk assessment based on interval type-2 fuzzy evidential reasoning method
Applied Soft Computing.
RFM-based repurchase behavior for customer classification and segmentation
Journal of Retailing and Consumer Services.
A Fuzzy Weighted Influence Non-linear Gauge System with Application to Advanced Technology Assessment at NASA
Expert Systems with Applications.
Towards a taxonomy-based preferred-customer model for suppliers in air cargo express service markets
Journal of Air Transport Management.
Bacterial colony algorithm with adaptive attribute learning strategy for feature selection in classification of customers for personalized recommendation
Neurocomputing.
Fuzzy sets
Information & Control.
A decision making model based on fuzzy inference to predict the impact of SCOR® indicators on customer perceived value
International Journal of Production Economics.
Cited by (8)
Optimal scheduling strategy for orderly charging and discharging of electric vehicles based on spatio-temporal characteristics
2023, Journal of Cleaner ProductionSoft computing for nonlinear risk assessment of complex socio-technical systems
2022, Expert Systems with ApplicationsA novel similarity measure in intuitionistic fuzzy sets and its applications
2022, Engineering Applications of Artificial IntelligenceCitation Excerpt :These theories have their own advantages in dealing with uncertain information. Therefore, researchers have applied those theories to various fields, such as decision making (Aydın et al., 2020; Gao et al., 2020a; Liao et al., 2021), target classification (Luo and Zhang, 2020; Geramian and Abraham, 2021; Fu et al., 2020), clustering problem (Egrioglu et al., 2020; Budwong et al., 2021; Ding et al., 2021) and other mixed areas (Lopes et al., 2020; Ma et al., 2021; Solanki et al., 2021; Gao et al., 2020b). Since IFS were proposed, similarity measure in the IFSs has been studied, such as Chen firstly defined the similarity measure between IFSs (Chen, 1995).
Using Fuzzy Logic to Make Decisions Based on Data From Customer Relationship Management Systems
2023, Advances in Science and Technology Research JournalNon-iterative Wagner-Hagras General Type-2 Mamdani Singleton Fuzzy Logic System Optimized by Central Composite Design in Quality Assurance by Image Processing
2023, Studies in Fuzziness and Soft Computing