Open Site Navigation

What You Need to Know About Predictive Analytics in HR

Predictive Analytics helps organizations leapfrog their competitors. Well-informed prediction models, to inform decision-making, are essential during times of change.



Assessments and analytics are increasingly used to predict employee behavior, engagement, performance, and risks of attrition. In this article, Eric Shepherd explains what predictive analytics is and how this will revolutionize our thinking.


What is Predictive Analytics?


We see examples of predictive analytics in our day-to-day lives. Credit scores, product recommendations, dating apps, and supply-chain management systems all use predictive analytics. Credit scores predict how likely we are to repay a debt. Finely tuned algorithms can predict products that we're like to purchase. Dating apps use personal data to connect people that could be a match. And supply-chain management systems can pre-order and pre-position products that are likely to be consumed. Unlike the 2002 movie Minority Report that uses "precogs" to predict crimes, we use machines to crunch data in making predictions.


The social media giant Facebook uses predictive analytics to predict which user is likely to buy a product and place ads, based on the data collected from their profiles and the posts. It is a technology that uses historical data and predicts future events or outcomes.


The question arises, how does it work? Predictive analytics are data-driven and use statistical techniques to forecast and analyze the outcome. A mathematical model or algorithm is created by analyzing the data collected to project potential futures. For organization’s leaders and HR teams, this enables them to build predictive models to understand the possible engagement and effectiveness of its workforce.


Practical Applications of Predictive Analytics


Managing a golf club provides a useful example. Playing golf is most enjoyable with perfect weather conditions. If the manager tracks reservations, course usage, weather predictions, and actual conditions, over time, there is a possibility of predicting bookings and usage. Armed with this information, a manager could prepare staff and equipment for the busy times and reduce the facility's capacity based on the data collected.

Does the manager have enough data to predict course usage reasonably? The answer is yes and using decision tree predictions can be more useful.


Using a Decision Tree for Predictions


A decision tree is a tree-like model of decisions and consequences; it is a powerful tool used in analytics, where each leaf, or node, denotes an attribute, and each branch indicates the potential effects.


In the case of the probability of golf course usage, let's think about two predictors to keep the model simple, i.e., sunny day vs. rainy day. The likelihood of playing golf diminishes if it rains. Whereas the possibility of playing golf increases when the day is sunny. In short, weather forecast data can be used to predict the likelihood of golfers playing golf. Weather forecasts are not entirely accurate and so keeping an eye on the weather is another useful predictor. The leaf is the data, such as its raining, and the brand is the consequence, we need less staff and less equipment. This simple case of managing a golf course helps us understand the principles. Now imagine that we have hundreds of data inputs, and we are trying to predict if someone will enjoy a specific movie. And now imagine that we're trying to provide that service for millions of individuals. The problem becomes exponentially more complicated. Machine learning, which is a technique for computers to produce their own statistical models, can help here by digesting numerous data points and making sense of the patterns to come up with predictions. When organizations apply it to their workforce, it can provide useful insights on engagement, risk of loss, and productivity.


Predictive Analytics in HR


Data stored by organizations in their Human Resource Information System (HRIS) can be used to develop predictive models for employee behaviors. HR predictive analytics can help formulate policies for employee well-being, engagement, and efficiency, and to predict the performance the organization is likely to achieve.


Here are some examples:


Averting Regretted Attrition


All organizations have a level of regretted attrition, that is losing employees that an organization truly regrets losing. An employee's tenure will depend upon their circumstances and the nature of the work. Employee turnover may be benchmarked against similar job roles and industry sectors. Regretted attrition has many negative consequences which include loss in revenue, added cost of hiring, unwanted distractions, and reduced productivity.


Employees that enjoy their work and fit in with the culture and have good managers are likely to stay with organizations. Calculating a "Flight Risk Score" using mathematical models can help to predict the possibility of a worker resigning. Over time these models will improve to help the organization understand additional predictors for attrition.


There are privacy-related issues with accessing flight risk scores. That means access to this data must be restricted on a need-to-know basis. The power of the flight risk score is to help a manager intervene to avoid a regretted resignation and give them a game plan in cases where an employee's departure is inevitable.


Hiring Right