Predictive analytics is full of tools and approaches enabling it to reveal key insights in almost any area. We already discussed the impact of Data Analytics in HR and we are delivering further.
Recent blog post by Toshi Takegushi, part of MathWorks team reveal in an interesting and comprehensive way how a predictive analytics model can be triggered on job-related data sets for scoring which employee is planning to quit its position. He relied on machine learning algorithms for predicting future events by utilizing historical data.
The data set was compiled of 10 different variables:
• Employee satisfaction level, scaling from 0 to 1
• Last evaluation, scaling from 0 to 1
• Average monthly hours
• Number of projects
• Salary, scaled as low, medium or high
• Earned promotion
• Work accidents
• Whether the employee has left
• Time spent in company in years
Takegushi modeled data set consisting 14999 people worked for this company for an unknown period. 24% of the employees quit the company within this unknown period. This is way above the healthy turnover of 5%, which is strongly varying in different regions and industries.
His insights show that if you break down the data by job function it seems that HR and accounting have low median satisfaction and high turnover percentage. The next step of the analytical solution was the measurement of high performers. Toshi Takegushi relied on last evaluation variable to found out who is better. The data work set the median at 0.72 so he presumed that anyone with a score of 0.8 or above is among the best employees. His data set shows clearly that highest risk of quitting the company is for people scored with 0.5 or 1.
On top of this is the experience variable. Data reveal that people who have worked for the company between 4 and 6 years show high probability to leave their positions. Then Takegushi added the satisfaction variable but he applied it to the high-risk groups.
His key finding was that not just low satisfied employees are leaving the company but also the ones with levels of satisfaction rated at 0.7 and above, and there is increased probability of quitting their jobs. The next step of his solution is to check the salary variable. By isolating high performers who are seasoned on their position it became visible that employees with lower or medium salaries are leaving their positions.
By applying predictive analytics, Takegushi created analytical algorithm forecasting which of the employees are going to leave, checked it out with holdout data and the outcome he got was 100% positive. All of the forecasted to leave had actually left the company and the opposite. The model utilized different variables at different weights with emphasize on the satisfaction, followed by the projects quantity, and the time spent at the company. This way it became visible that employeeswho got a promotion in the last 5 years are less likely to leave.
Where is the HR analytics benefit of this all?
Such a model easily predicts when an employee is going to leave his company. HR experts might trigger or not action in order to keep this particular professional within the organization. The analytical solution showed clearly that satisfaction, promotion, and salary are key factors in retaining valuable employees.
Take a look at a live demo of this particular predictive model.