Job performance prediction in a call center using a naive Bayes classifier
Highlights
► We model a naive Bayesian classifier to predict the job performance of sales agents of a Call center. ► We show that socio-demographic attributes are not suitable for predicting performance. ► Daily operational records were used to predict production of sales agents with satisfactory results. ► Inference made on the naive Bayesian network permits to establish particular minimum conditions of operation of the sales agent to remain in the firm.
Introduction
Employee turnover has been a fertile ground for numerous studies that deal with, from a theoretical perspective and empirical, the causes of both, voluntary and involuntary turnover, as well as those that explain the tenure of employees. Particularly in call centers, employee turnover is a problem related to high turnover rates, which represents an important economic cost such as advertising, recruitment, testing, training and supervision at work to achieve, after a certain time, the employee’s competence as to achieve adequate levels of productivity (Hillmer et al., 2004, Robb, 2002). Sums of money are spent in recruiting and training people, but unfortunately, a large percentage of them leave the firm before reaching some kind of reasonable performance. According to Robb (2002), this is an indication that the recruitment and selection activities are not confronting or preventing the central problem: the high turnover rate. This problem causes, according to Jackson (2009), that 70% of the costs of operating a call center are related to personnel management. Employee turnover contributes to a high ratio of staff costs to operating costs.
Call centers represent an exceptional case of the service industry and labor of customer contact, in which the mechanization of work has been installed as a form of taylorism to streamline production. Usually, this model of mass production is characterized by requiring instrumental competences, little opportunity for making decisions, and learning is limited to the repetition of tasks. In this case, it is assumed that the work is designed to turnover-proof where workers are like replacement parts (Batt & Moynihan, 2002). From this perspective, one might assume that the management style of this model naturally attracts high turnover. Just as it happens in Chile, the problem is exacerbated when market conditions remain competitive in the industry resulting in the loss of thousands of employees Concha (2010). The pressure to stay competitive in the industry, leads to call centers to put maximum effort in operational efficiency, using technology to centralize the calls at one location and where labor costs are lower and more labor flexibility (Buchanan, 2005).
Call center executives often are subjected to demands that can lead to stress and frustration (McDonald, 1998, Ruyter et al., 2001). In many cases, executives must deal with aggressive clients, be subject to constant on-line monitoring and have low flexibility and chance to operate under its own discretion. Under this scenario, it is relevant the ability of the call center to select individuals who may have a good performance and a longer service time. The selection process should identify certain characteristics of a personal nature and personality that are consistent with the organization and working environment of the call center itself (Adorno, 2010). The process should try to avoid a poor fit between the individual’s job expectations and the expectations that the organization has to it. The firm must pay attention and effort in the recruitment and selection process from the perspective of where to look in relation to the skills and knowledge of the candidate, so that it can meet the requirements of the business (Jackson, 2009, Birnbaum and Somers, 1997).
This study aims to assess individuals during her/his test period using a naive Bayes classifier to discriminate between cases that achieve or not the minimum performance, required by the firm to remain in it, using demographics and operational attributes. The structure of this paper is as follow: First, we review recent studies that address the problem of turnover and employee selection. Second, we present a general model of individual performance and the methodology used to evaluate it. Third, we presented the results of simulations and predictive capacity of the model and, finally the conclusions.
Section snippets
Related work
Recruitment has been an important aspect of research in human resource and organizational psychology. The discovery of patterns and relationships in the personal data characteristics of people, can help to predict the behavior of individuals in terms of their performance. Research from these areas have produced important developments concerning with the relationship between personality characteristics of individuals and their expectations, satisfaction and job performance (Borman and Motowildo,
Background
The work presented in this paper was carried out at a Call Center located in Santiago, Chile. It is dedicated to meet requirements of its institutional clients in telemarketing and sales activities of banking, retail and telecommunications business. It has an average of 500 sales agents per month. The campaign is the core activity of the call center focused on making a sale through the telephone channel services provided by institutional clients and offered to consumers for an indefinite period
Methodology
This research was developed following the CRISP-DM model proposed by Chapman et al. (2009). CRISP-DM model provides a generic guide to develop a data mining project life cycle. Data is collected from the daily operations of the sales activities of agents and data bases of the Human Resource Department. This information regarding personal data and operational performance of sales agents, were used to investigate the attributes that explain the production. The dependent or class variable Y is a
Results
A first simulation using socio-demographics attributes (age, gender, marital status, education level, socioeconomic level of the county of residence and experience) was carried out to discriminate between those who achieve above or below the minimum production to remain during the insertion phase. It was expected, that a married person had greater incentive to perform better than a single person, or that an older person should performer better than a young person. One might expect that a person
Conclusions
Taking a subset of personal attributes of an individual, is insufficient to predict the performance at t = 0. It is possible that in jobs with low specialization and high routine, as in the case of sales and telemarketing activities in the call center, attributes such as level of education, age or marital status have little influence in performance and turnover intentions, as opposed to other professions with greater degree of specialization. The operational attributes proved to be a better
Acknowledgments
The authors would thank Alejandro Ventura, CEO of the Call Center Hispanicglobal for their willingness and support during the implementation of the model. We also thank Margarita Quiroz who belongs to the systems department of the Call Center for her support in obtaining the data bases. This work was supported financially by the Universidad Adolfo Ibáñez and the Universidad de Valparaíso.
References (26)
- et al.
Data mining to improve personnel selection and enhace human capital: A case study in high-technology industry
Expert System with Applications
(2008) An introduction to ROC analysis
Pattern Recognition Letters
(2006)- et al.
On using direct measures of met expectations: A methodological note
Journal of Management
(1995) - et al.
Decision support system for conflict diagnosis in personnel selection
Information & Management
(1999) - Adorno, A. J. (2010). Understanding and combating employee turnover in call centers....
- et al.
The viability of alternative call center production models
Human Resource Management Journal
(2002) - et al.
Fitting job performance into turnover model: An examination of the form of the job performance-turnover relationship and path model
Journal of Management
(1997) - et al.
Task performance and contextual performance: The meaning for personnel selection research
Human Performance
(1997) Call centers: Ringing in profits
Export Wise
(2005)- Chapman, P., Clinton, J., Kerber, R., Khabaza, T., Reinartz, T., Shearer, C., et al. (2009). Crisp-DM 1.0: Step-by-step...
Data mining for selection of insurance sales agents
Expert System
A fuzzy MCDM approach to personnel selection
Expert Systems with Applications
Cited by (54)
Responsible machine learning for United States Air Force pilot candidate selection
2024, Decision Support SystemsPredicting success in United States Air Force pilot training using machine learning techniques
2022, Socio-Economic Planning SciencesCitation Excerpt :In adopting this perspective, our research extends the application of machine learning in human resource and educational tasks to the military environment. In the civilian sector, machine learning has been applied to hiring practices [8], job performance prediction [9], employee retention [10], and recruitment [11]. [12,13] provide thorough literature reviews in this regard.
Employees recruitment: A prescriptive analytics approach via machine learning and mathematical programming
2020, Decision Support SystemsCitation Excerpt :Some of these studies examined the post-hire prediction of turnover or performance with predictors collected over the employment period [8,16–23]. Note that the prediction of turnover or performance using post-hire data (such as absenteeism, punctuality and performance reviews) may be useful as part of some retention activities but may lead to a late discovery of recruitment errors and may often be too late to act upon [8,24]. In contrast, the potential benefit of the early pre-hire foresight of longer-term employee success may be much higher, saving more financial and social costs.
A Multi-criteria Decision Making Model Integrated with Speech Analytics for the Performance Evaluation of Agents in Express Delivery Industry
2024, Lecture Notes in Mechanical EngineeringAn empirical investigation on the effect of applying artificial intelligence tools in human resource analytics
2023, Disruptive Artificial Intelligence and Sustainable Human Resource Management: Impacts and Innovations - The Future of HR