Madhusudhan_End Exam_Q&A

1. The Voluntary Terminations of employees are 88. The top 4 reasons are: I. Another Position – 20 II. Unhappy-14 III. More money – 11 IV. Career Change – 9 2. 14 Female employees from production department have left because of “Another Position” as a stated reason. 3. 11 people are joining in near future. 4. The top 4 most expensive recruitment channels on the basis of cost per employee recruited are: I. CareerBuilder – 7790 II. Pay per Click – 1323 III. MBTA ads – 645.88 IV. On-Campus Recruiting – 625 5. Total no. of employees joined through employee referral = 31 Total no. of employee actively working = 24 Retention rate = (24/31) *100 = 77.41% 6. Based on the data available and the scenario provided in the previous questions, Dependent Variable that I choose is Days employed and the independent variables will be employee source, department, sex, reason for term, cost per employee calculated field, employee name, employee id, married id and manager. Main reason for choosing these variables is to identify the voluntary terminations in the productions department and the terminations happening through various employee sources. Linear regression dependent variable should be a measure so “Days employed” is chosen. 7. For Logistic regression the dependent variable would be employee status because the objective is to find the pattern of people based on employee status. Further drilling down would provide the reasons of people leaving or staying in organization. 8. If we have to use CART, then we should choose Department as dependent variable when using the CART model. CART algorithm is the process of answering the sequence of questions in a hierarchical basis. So, for predicting the better results we start analyzing from the top level like department and start drilling down for the employment status and reason. 9. Association Rules: Rule 1: {Manager Name = Simon Roup; MartialDesc = Single; Performance Score = Fully Meets} => {Employment Status = Terminated for cause} Lift = 19.8571 (Quite high as compared with other rules) Interpretation: The probability an employee being terminated for cause is very high when manager is “Simon Roup”, MartialDesc = “Single” and when the performance score was fully met. Rule 2: {Manager Name = David Stanley; MartialDesc = Divorced} => {Employment Status = Voluntary Terminated} Lift = 3.394 Interpretation: The probability an employee voluntarily terminated is high when their manager is David Stanley and the MartialDesc is divorced. Rule 3: {Manager Name = Webster Butler; Performance Score = 90-day meets} => {Employment Status = Voluntarily Terminated} Lift = 3.3494 Interpretation: The probability an employee does voluntary terminated is high when their manager is Webster Butler and when the performance score was just met for 90 days. Rule 4: {MartialDesc = Divorced; Performance Score = N/A – too early to review} => {Employment Status = Voluntarily Terminated} Lift = 3.3494 Interpretation: The probability an employee prefers for voluntary termination is when the employee is divorced, and her performance appraisal cycle would have not opened as she would have joined recently. Rule 5: {Manager Name = Brannon Miller; Position = Production Technician II; Sex = Male} => {Employment Status = Voluntary Terminated} Lift = 3.3494 Interpretation: The probability an employee prefers for voluntary termination is when the manager is Brannon Miller, they are not satisfied with their position Production Technician II and mainly male employees.