Your Highest Cost: Employee Burn Out

Burnout rate prediction model and decision making dashboard

How is the COVID-19 crisis reshaping remote working? What is the best way to enhance employee productivity? Remote work raises a wide range of challenges for employees and employers. Employees are struggling to find the best work-life balance, working remotely, while managers envision employees lying on their couches eating Pringles and watching the James Corden show.

Happy and healthy employees are more productive at work, and in turn, help the business flourish profoundly.

However, since working from home has become the new strange normal, over 69% of employees have been showing burnout symptoms globally (source). On the occasion of World Mental Health Day (October 10th), HackerEarth shared a challenge to build a machine learning model that predicts the employee burnout rate based on numerous factors.

The aim of this project is not only to explore the dataset and define a model for burnout rate prediction but also to get inspired on how to use the result to create an interactive decision-making dashboard.

If first, you are interested in the analysis techniques (dataset encoding, imputation, prediction model), please see the Github page of the project.

What aspects are highly correlated with burnout rate?

From the heatmap below, it can be observed that Mental Fatigue Score, Resource Allocation (working hours) and Designation variables are highly correlated with employee burnout rate. Working from home (WFH) setup availability has an opposite, positive effect on the employee burn-rate.

How does remote work affect burnout rate?

The average burnout rate is 0.396 in the home office setup availability group and 0.518 in the office group (burnout rate range is 0–1). The resource allocations (working hours) are less in the home office setup availability group (mean 3.947 and 5.118). It‘s not clear that this difference is due to the more efficient performance, less allocated workload or the first sign of decreased motivation.

How well can we predict an individual’s burnout rate?

In this case, a simple linear regression was used to estimate the relationship between the variables and the burnout rate. The R-squared value (the proportion of variance in the dependent variable that can be explained by the independent variables) was 0.92, which is a strong level.

We created our predictive model. Move to the next step, using the prediction to improve business performance.

How much does burnout cost your business?

From a business standpoint, the cost of employee burnout can be significant (sickness absence, presenteeism, higher turnover, company reputation). It has a negative effect on nearly every aspect of a business. The results show that every $1 of expenditure in promotion and prevention program generates net economic benefits over a one-year period of up to $13.69 (source).

Burnout is particularly high among employees at Google and Facebook: 79% of Google employees who responded to a survey say they’re more burned out than before (source), and 81% of Facebook employees said the same.

We will use ActiveGraf to showcase our scenarios. I created an interactive PowerPoint presentation to highlight the impact of a corporate prevention program. My presentation is connected with an Excel file, that contains the business assumptions, and the model runs the prediction script each time we set a new goal. In this way, we empower everyone to be able to run any scenario in real-time, based on this machine learning predictive model.

Let’s see why the prevention of burnout can be so important for Google. The stress-related cost can be somewhere around 19% of the total human resource cost. By triggering a 25% reduction in Mental Fatigue Score (our strongest predictor of burnout rate) we can reach a 0.1 reduction in the burnout rate. This results in an 8% reduction in human resources cost which means a $ 751 mln savings annually.

And how you will prevent employee burnout?