Applications of analytics in HR: Using predictive analytics techniques
Analytics has not traditionally been applied to the field of Human Resources largely because analytics requires a solid understanding in the foundations of Linear Algebra, Statistics, and Calculus. However, there are organizations that have begun competing in predictive analytics in the field of HR. In this post I will be discussing the applications of predictive analytics and how it positively impacts the organization.
Analytics has not traditionally been applied to the field of Human Resources. A major part of that is because math has never been known to be a strong suit for Human Resources professionals (Davenport & Harris, 2017). Analytics requires a solid understanding in the foundations of Linear Algebra, Statistics, and Calculus. Therefore, the usage of analytics has mostly been descriptive and thus the entire potential of predictive analytics is left untapped.
Most organizations engage in simple reporting to pull data on employee statistics. Advanced analytics involve the use of predictive analytics to forecast the future (Jones, 2014). Predictive analytics has more applications. I will discuss the applications of predictive analytics and how it positively impacts the organization.
There are organizations that have begun competing in predictive analytics in the field of HR (Davenport & Harris, 2017). In general, being able to compete on HR analytics means that a huge investment made by an organization can be more effective, efficient, and lower cost than their competitors.
Analytics can be used to support the main goals of HR which are to attract, retain, motivate, and develop top talent. Currently, it’s common practice to use intuition and guesswork to hire and attract employees. Most organizations still use a resume, phone screen, and in person interview to identify top talent. These techniques have as good of a chance as the scouts in Moneyball (De Luca, M., Horovitz, R., Pitt, B (Producers), & Miller, B. (Director). 2011) to select top baseball players using their experience and intuition.
Logistic regression is a predictive analytics tool that can improve how organizations attract employees. Logistic regression allows you to output values into categories (Brownlee, 2016). One way to apply this to HR recruiting is to identify all of the recruits that ended up being top talent and then figuring out what factors led to the recruit being a top talent. An example could be that recruits that went to Northwestern University are more likely to end up being a top talent so this could be one of the predictors that categorize a recruit as a top talent or not.
In terms of retaining top talent, the chances are just as bad because HR and managers are usually reactive rather than proactive. A common scenario is where an employee starts applying to other companies because they aren’t happy and when they finally get an offer, the company is forced to react by offering to match it or pay even more to keep this employee. It would be better to know when an employee is thinking about leaving or to know what programs work to keep employees.
A tool that predictive analytics can use is a machine learning algorithm called K-nearest neighbors (Navlani, 2018). This is a simple approach to classify current employees into their risk level of leaving the company. Organizations can look at historical employees that left and input all of their employee data to set up a model that classifies when employees are going to leave. This model can then identify which employees are at risk for leaving so that proactive measures can be taken to keep the employee.
Another challenging process is forecasting headcount and labor cost. A common method to forecast labor cost and headcount is by looking at the numbers in the prior year and then adjusting based on a best guess. Usually these forecasts are very far off from what actually happens.
Time series forecasting can be used (Liu, Wei, & Zhang, 2017) for headcount and labor cost forecasting. This technique begins with looking at historical data, but uses various mathematical methods to forecast the data. The benefit of having better forecasts mean being able to plan your resources better. Similar to a previous idea discussed, finance can allocate resources better if you aren’t over or under forecasting your estimates.
We began by talking about how analytics has been underutilized in the field of HR, then moved on to specific challenges in HR, and then moved onto methods that analytics can use to address those challenges. Now we will talk about the specific benefits to businesses that embrace these methods.
Information is power (Davenport & Harris, 2017). Organizations that have more information about the future can plan better. In addition to better planning, organizations have less risk in the decisions they make. One interesting way of measuring the impact of analytics on business outcomes is using causal pathway modeling similar to how Sears modeled the impact of employee attitudes on revenue (Rucci, Kirn, & Quinn, 1998). In the end, analytics can help HR functions save on cost, improve effectiveness, and provide the ability to plan better.
Davenport, T.H., & Harris, J.G. (2017). Competing on analytics: The New Science of Winning. Boston, MA: Harvard Business School Publishing Corporation.
Liu, T., Wei, H., Zhang, C., & Zhang, K. (2017). Time series forecasting based on wavelet decomposition and feature extraction. Neural Computing & Applications, 28, S183-S195. doi:http://dx.doi.org/10.1007/s00521-016-2306-8
Jones, K. (2014). Conquering HR analytics: Do you need a rocket scientist or a crystal ball?. Workforce Solutions Review, July.
Rucci, Q.J., Kirn, S.P., & Quinn, R.T. (1998). The employee-customer-profit chain at sears. Harvard Business Review, January-February.
Navlani, A. (2018). KNN classification using scikit-learn. Retrieved from https://www.datacamp.com/community/tutorials/k-nearest-neighbor-classification-scikit-learn
Brownlee, J. (2016). What is time series forecasting?. Retrieved from https://machinelearningmastery.com/time-series-forecasting/
Brownlee, J. (2016). Logistic Regression for Machine Learning. Retrieved from https://machinelearningmastery.com/logistic-regression-for-machine-learning/
De Luca, M., Horovitz, R., Pitt, B (Producers), & Miller, B. (Director). (2011). Moneyball [Motion picture]. United States: Scott Rudin Productions.