team b4- employee attrition analysis
TRANSCRIPT
Employee Attrition Analysis
Cohort B Team 4 Suraj Shah, Mengzhen Li, Xi Gong
Background
• SanDisk• Digital storage leader• Portable flash drives
• Problem: High Employee Attrition Rate
Employee Attrition• Loss of employees: retirement, resignation• Impact: cost to company
• training new and old employees• reputation
Data• Publicly available• CrowdANALYTIX• Unstructured
Objective
• Identify key indicators of employee attrition• Analyse patterns that could be indicative of ‘risk of
leaving’
• Example:• Software positions over 1.5 years: “high” risk of leaving• Manager positions over 3 years: “low” risk of leaving
Cleaning the Data
Structured Data
Unstructured Data
Identify SanDisk Employees
• Example:• Green: Exp1-Dummy=0, Exp2-Dummy=1, it means that
currently not working with SanDisk but previously were.
• Dummy variables : SanDisk = 1, Other = 0
Group Employees with Different Attributes
Percent of Males who left SanDisk 245/383 = 63.9%Percent of Females who left SanDisk 116/171 = 67.8%Percent of Missing Genders who left SanDisk 41/69 = 59.4%
Distribution Analysis - SPSS
Clustering
K-Means Kohonen
Discovering Attrition Patterns
• Tableau: Bubble Chart, Bar Chart• Inputs: Experience 2, Program 1, Average Duration-
Experience• Subgroups: • Experience 2: Engineering, IT, Sales & Marketing,
Operations, HR, etc.• Program 1: Computer Science, Electrical Engineering,
Marketing, etc.
Insights on Females – Cluster 1
Insights on Males – Cluster 2
Conclusion
• SanDisk can match any new joining employee’s characteristics with those in the visualization and make an intelligent guess as to how long an employee would stay
References
• www.CrowdANALYTIX.com• www.SanDisk.com
Thank You!