Gregg Gordon of Kronos outlines how machine learning can produce data models that can help HR make better decisions
It’s hard to find an industry that is not being impacted by machine learning applications. While often highlighted in finance or operations, machine learning will also have a significant impact on the HR industry. The following three examples show how machine learning is being applied today and how it can be used in the future to aid recruitment decisions and improve employee retention.
While machine learning is very good at detecting subtle patterns over time, with a limited number of variables - it is also a versatile tool.
Machine learning is capable of analysing many different sources of data to find relationships that would otherwise be impossible for an individual to uncover through experience or simple spreadsheet analysis. An emerging area where this is being applied is within the HR industry which is using it to predict employee outcomes.
Think of the many unknown reasons behind why some people are promoted or others leave a job. By ingesting the many variables such as tenure, wage, job title, time in current role, and education; machine learning algorithms can find the drivers of certain outcomes that would otherwise be lost in a sea of data.
By training these algorithms on the drivers and outcomes of thousands of previous employees, the indicators of success can be identified and acted upon. While this sounds good, it can also be a challenging proposition to pull in all the variables that affect turnover and success.
They can be different in different parts of the company and what occurred in the past may not be present today - affecting the results. These are early days for predicting an individual’s trajectory within the organisation and HR should think about using these predictions as directional indicators as opposed to absolutes as the algorithms mature.
One of the more difficult challenges in a company is ensuring that supervisors are treating their employees fairly while maintaining compliance with regulatory and corporate policies. It’s difficult because when supervisors are not treating their employees fairly, or according to policy, it may not be reported for fear of supervisor retribution.
It could be that employees are not aware a failure is occurring but end up resenting the job. Today, machine learning based applications can detect unusual but typically undetectable patterns in data, based on a supervisor’s actions, relative to time or scheduling inputs. These unusual patterns could indicate wider operational issues such as inaccurate labour forecasts.
For example, if a supervisor was faced with an inaccurate labour forecast, they may spend extra time manually editing a schedule to correct for that error but inadvertently cause hardship on the schedules of those employees affected. These situations often go unnoticed by management because each event is too small to register on traditional analysis. But to the individual who is impacted, it is significant and in cumulative effect to the company, it can explain why performance isn’t as expected.
Machine learning is not limited to assisting HR professionals in understanding employee behaviors as described in the previous two examples. It can also be applied more simply to administrative tasks that make employees’ lives better. To imagine what that could look like, let’s turn to the music industry. The music industry is learning how to live in a digitised world and musicians are dependent on maintaining a relationship with its customers to expand their revenues beyond the royalties earned from low cost music subscriptions.
What celebrity musicians have found is that they can provide personal information and responses to their fans’ questions online. From signing up fans to newsletters to providing information to featuring upcoming concerts; these musicians have employed the use of machine learning backed chatbots.
These chatbots don’t try and imitate the musician, rather they provide a unique experience that enhances the interaction between the fan and the franchise. This application could easily be envisioned as the next generation of HR self-service. By guiding employees through complex transactions, retrieving basic personal information or simply answering questions about benefit options and providers, chatbots will enhance the experience of employees when they interact with HR.
As these examples show, machine learning will make HR more effective and efficient. Applying machine learning to HR operations will eliminate some of the current roles that a HR professional performs when it comes to organising information or answering questions. In other cases it will uncover the answers to the perennial questions regarding why some employees leave and others excel in their jobs. It will allow HR professionals to begin to make data driven decisions and deliver innovation in the areas of recruitment, organisational design, workforce management, and employee engagement.
Gregg Gordon is the Vice President, Data Science Practice at Kronos
While machine learning is very good at detecting subtle patterns over time, with a limited number of variables - it is also a versatile tool.
Machine learning is capable of analysing many different sources of data to find relationships that would otherwise be impossible for an individual to uncover through experience or simple spreadsheet analysis. An emerging area where this is being applied is within the HR industry which is using it to predict employee outcomes.
Think of the many unknown reasons behind why some people are promoted or others leave a job. By ingesting the many variables such as tenure, wage, job title, time in current role, and education; machine learning algorithms can find the drivers of certain outcomes that would otherwise be lost in a sea of data.
By training these algorithms on the drivers and outcomes of thousands of previous employees, the indicators of success can be identified and acted upon. While this sounds good, it can also be a challenging proposition to pull in all the variables that affect turnover and success.
They can be different in different parts of the company and what occurred in the past may not be present today - affecting the results. These are early days for predicting an individual’s trajectory within the organisation and HR should think about using these predictions as directional indicators as opposed to absolutes as the algorithms mature.
One of the more difficult challenges in a company is ensuring that supervisors are treating their employees fairly while maintaining compliance with regulatory and corporate policies. It’s difficult because when supervisors are not treating their employees fairly, or according to policy, it may not be reported for fear of supervisor retribution.
It could be that employees are not aware a failure is occurring but end up resenting the job. Today, machine learning based applications can detect unusual but typically undetectable patterns in data, based on a supervisor’s actions, relative to time or scheduling inputs. These unusual patterns could indicate wider operational issues such as inaccurate labour forecasts.
For example, if a supervisor was faced with an inaccurate labour forecast, they may spend extra time manually editing a schedule to correct for that error but inadvertently cause hardship on the schedules of those employees affected. These situations often go unnoticed by management because each event is too small to register on traditional analysis. But to the individual who is impacted, it is significant and in cumulative effect to the company, it can explain why performance isn’t as expected.
Machine learning is not limited to assisting HR professionals in understanding employee behaviors as described in the previous two examples. It can also be applied more simply to administrative tasks that make employees’ lives better. To imagine what that could look like, let’s turn to the music industry. The music industry is learning how to live in a digitised world and musicians are dependent on maintaining a relationship with its customers to expand their revenues beyond the royalties earned from low cost music subscriptions.
What celebrity musicians have found is that they can provide personal information and responses to their fans’ questions online. From signing up fans to newsletters to providing information to featuring upcoming concerts; these musicians have employed the use of machine learning backed chatbots.
These chatbots don’t try and imitate the musician, rather they provide a unique experience that enhances the interaction between the fan and the franchise. This application could easily be envisioned as the next generation of HR self-service. By guiding employees through complex transactions, retrieving basic personal information or simply answering questions about benefit options and providers, chatbots will enhance the experience of employees when they interact with HR.
As these examples show, machine learning will make HR more effective and efficient. Applying machine learning to HR operations will eliminate some of the current roles that a HR professional performs when it comes to organising information or answering questions. In other cases it will uncover the answers to the perennial questions regarding why some employees leave and others excel in their jobs. It will allow HR professionals to begin to make data driven decisions and deliver innovation in the areas of recruitment, organisational design, workforce management, and employee engagement.
Gregg Gordon is the Vice President, Data Science Practice at Kronos