Predict employee attrition by using predictive analytics

Abstract

Purpose Research questions that this paper attempts to answer are – do the features in general email communication have any significance to a teaching faculty member leaving the business school? Do the sentiments expressed in email communication have any significance to a teaching faculty member leaving the business school? Do the stages mentioned in the transtheoretical model have any relevance to the email behaviour of an individual when he or she goes through the decision process leading to the decision to quit? The purpose of this paper is to study email patterns and use predictive analytics to correlate with the real-world situation of leaving the business school. Design/methodology/approach The email repository (2010–2017) of 126 teaching faculty members who were associated with a business school as full-time faculty members is the data set that was used for the research. Of the 126 teaching faculty members, 42 had left the business school during this time frame. Correlation analysis, word count analysis and sentiment analysis were executed using “R” programming, and sentiment “R” package was used to understand the sentiment and its association in leaving the business school. From the email repository, a rich feature set of data was extracted for correlation analysis to discover the features which had strong correlation with the faculty member leaving the business school. The research also used data-logging tools to extract aggregated statistics for word frequency counts and sentiment features. Findings Those faculty members who decide to leave are involved more in external communication and less in internal communications. Also, those who decide to leave initiate fewer email conversations and opt to forward emails to colleagues. Correlation analysis shows that negative sentiment goes down, as faculty members leave the organisation and this is in contrary to the existing review of literature. The research also shows that the triggering point or the intention to leave is positively correlated to the downward swing of the emotional valence (positive sentiment). A number of email features have shown change in patterns which are correlated to a faculty member quitting the business school. Research limitations/implications Faculty members of only one business school have been considered and this is primary due to cost, privacy and complexities involved in procuring and handling the data. Also, the reasons for exhibiting the sentiments and their root cause have not been studied. Also the designation, roles and responsibilities of faculty members have not been taken into consideration. Practical implications Business schools all over India always have a challenge to recruit good faculty members who can take up research activities, teach and also shoulder administrative responsibilities. Retaining faculty members and keeping attrition levels low will help business schools to maintain the standards of excellence that they aspire. This research is immensely useful for business school, which can use email analytics in predicting the intention of the faculty members leaving their business school. Originality/value Although past studies have studied attrition, this study uses predictive analytics and maps it to the intention to quit. This study helps business schools to predict the chance of faculty members leaving the business school which is of immense value, as appropriate measures can be taken to retain and restrict attrition.

Intelligent Employee Retention System for Attrition Rate Analysis and Churn Prediction

“…The implications from existing studies (Raman et al,2019;El-Rayes, Fang, Smith, and Taylor,2020) indicate that the primary factors that influence employee churn are: 'Employee_ Satisfaction', 'Appraisal_rating', 'Employee_CTC_level', and workload ('Number of projects/tasks assigned per quarter').…”

“…• Appraisal_rating: The metric that measures how well the employee performs a task within the time and budget constraints in the form of Appraisal_rating is also an essential factor for the organization provided out of 2 (Mohr et al,2012;Keegan and Hartog, 2019). • Employee_CTC_level: The level of annual remuneration of the employee (measured in Cost to Company i.e., CTC) and whether he is promoted or not commensurate to his performance evaluation score also in-directly impacts 'Employee_Satisfaction' and, in turn, 'Turnover_status' (Ajit,2016;Ogbonnaya et al,2017;Ma et al,2019;Raman et al,2019;Mohr et al,2012;El-Rayes, Fang, Smith, & Taylor,2020). The 'Employee_CTC_level' is classified into low, medium, and high, which is recoded in the dataset for further analysis as 0, 1, and 2.…”

“…. Raman et al, (2019) The objective of this paper is to study the factors that drive employee retention and reduce attrition.…”

The role of predictive analytics to explain the employability of management graduates

“…The literature indicates that many Indian B-Schools focus on increasing the salary packages of graduates. While analytics has been used by a few Indian B-Schools to predict the faculty attrition, using it for predicting placement is not indicated in the literature (Raman et al , 2019). There is significant literature that indicate the need to restructure and improve management education (Bennis and O'Toole, 2005; Donaldson, 2002; Emiliani, 2006; Etzioni, 2002; Ferraro et al , 2005; Ghoshal, 2003; Hamel, 2009; Mintzberg et al , 2002; Pfeffer and Fong, 2002; Tsurumi, 2005).…”

Analyzing the influence of job satisfaction on hostage employee loyalty in Isfahan Province Gas Company

“…From this point of view, this study has contributed to internal benchmarking in addition to the external benchmarking addressed by Fawcett et al (2004). Since the focus of this paper was on a special category of employees, i.e., hostage category in which, employee satisfaction was not high, this paper has also contributed to predicting employee attrition (Raman et al, 2018) as a factor enforcing employees to leave the organization. The demographic distribution of respondents indicated that 52 percent of employees were at the age of 40 and higher.…”