In  Few algorithms were used
In 1 Few algorithms were used, and their performance were analyzed on the dataset and it was observed that the random forest algorithm attained the highest accuracy and sensitivity, followed by decision tree algorithm. It can be realized that the random forest algorithm, an ensemble model that incorporates many decision trees, is evidently better at predicting the employee attrition result in comparison to the other machine learning methodologies. Linear SVM was identified to give lowest results among the models used. Kernel SVM or SVM tunings might be used to improve the accuracy of the existing SVM model.
2 analyzed the reasons for attrition by considering the cases of staffs from 3 major IT firms in Chennai using SPSS and the outcome showed that the primary two reasons for employees to leave an organization are poor appraisal patterns and lack of career advancements in their previous workplace. It has been validated that employees tend to get highly unsatisfactory when the appraisal and recognition they receive is not in line with their work. Heavy workload and Gender based discriminations have also been observed to be a deciding factor when it comes to workers attrition. Conflict with supervisors, not in a constructive way stands third here as a hostile environment between subordinates and supervisors disrupt the quality and smoothness in the work environment.
Other notable factors that impact attrition include overtime, low package, sparse motivation and acknowledgement, unreal workload.
3 suggested a clustering method for voluntary attrition estimation as the dataset used had confidentiality restrictions and limited worker information. Due to this constraint, job category level is used in this unsupervised clustering model, instead of individual level.
The hypothesis of this work deals with how various factors of the given dataset affect the one dependent variable (attrition) of the employee.
A. The Bridge between Employee attrition & Customer attrition – How it relates to Customer Relationship Management
4 Relates employee attrition to the customer attrition where a customer ends the services he/she received from a company voluntarily. This article talks in detail about the customer lifetime value model and employee value model, drafting them similarities and using the customer model to predict and describe the employee attrition. Few improvisations like strengthening the services provided to customers, improvement in offering discounts and benefits, robust and reliable customer service have all fortified customer-company relationships and leads to a low customer churn rate. High competitions have resulted in customers swaying from their loyalty to the original company. As it is too late once the customer switches to a different competitor company, pro-active actions are necessary to identify any potential cases of customer churn and take the essential steps to address the same. Machine learning and predictive models empower analysts to identify and extrapolate the cases in such scenarios so that the company can then take the preventive measures to hold on to their customers. Similar is the case for employee attrition where statistical models can be used to identify cardinal factors that lead to attrition and gives a leverage to the higher management in deciding the next step. However, it also needs to be understood that the reasons for employee churn is more intricate and involves many human related components in it whereas customer churn will mostly be mapped to simple issues and shortcomings.
An employee value model is created in this research based on various attributes like employee onsite and offsite durations, employee billability with the clients and the criticality of the project that the employee was a part of. The reason behind this model is to address the complexity in retaining all the workers who are predicted to churn as it is not possible to retain all. Hence, this novel model determines the highly valuable workers among the entire churn list and then emphasis is laid on those workers to retain them.
In customer relationship management, similar models are developed i.e., a customer Value Model to identify high prospect customers and offer them better services at lower price to retain them 6. It explains features like
1. customer value over lifetime.
2. Length of Service explaining the customer churn over a time.
3. The discount factor that estimates the value of the customer in terms of the profit attained in the future that contribute to this model.
5 has used ANOVA to understand few factors and their effect on employee attrition. In the research, two hypotheses were put forward and verified. Once was related to the health issues and how it might impact the workers at different shifts in a company. The output bolstered the hypothesis and concluded that there is no statistically significant difference among the health-related opinions among employees of various shifts. In other words, there was no major difference reported by employees of various shifts, in relation to health. Secondly, a Chi square test of independence was performed between skill utilization and years of experience of a worker. It was found that experience had an impact on skill utilization, where the skills of workers with high experience were under used.
Few interesting inferences were made in the above work where it was decoded that there was a huge number of workers ages below 20 who quit jobs due to high work pressure, inadequacy to adapt to the new and challenging corporate environment, the rate was high especially among females. Another factor the caused these young workers to quit jobs was the shift structure in the corporate firms, BPOs and call centers to be more precise as they were not able to accustom themselves to the work schedule. There were similar observations with previous papers in terms of salary and better opportunities as the rate of attrition was high among employees aged between 20 and 25 owing to these reasons. Also, monotonous job caused boredom among the workers between 21-25 that caused them to switch jobs. Among females again, there has been considerable attrition among 20-30 years of age due to family reasons, pregnancy etc.,