Analytics is gaining power over our lives as algorithms are making more decisions. How do we know those decisions are right?
Ten years have passed since Clive Humby, a UK mathematician, said that data is the new oil. Like crude that must be changed into gas or chemicals to create valuable entities, data must be broken down and analyzed for it to have value.
Data is only meaningful as it relates to the marketplace. By itself it does not provide insight, and facts alone are not enough for good decision making. One key insight changed everything for Starbucks when Executive Chairman Howard Schultz realized that “we serve people coffee; we don’t serve coffee to people.” Starbucks knows they are not about coffee, but about the experience of enjoying it. This concept drove all the decisions made by the company leading to its global success.
Good decisions depend on good data
But decision making in today’s data driven culture is no simple matter even if you are very clear on your company’s vision and purpose. Good decisions depend on good data, but with data coming in from many different sources in multiple formats organized by a variety of criteria, working with clean data sets is an increasingly difficult process. In addition, data management is becoming more and more automated.
As the use of analytics gains power over our lives, how do we know automated decisions are right? The SAP Analytics portfolio announced at SAPPHIRE NOW this week is designed to help.
It Takes Trust
Typically, as predictive or optimization models are introduced to support human decision making, it begins with a suggested decision. This allows the decision maker to review options, but still make the final decision. As decisions prove to be the right ones for the business over time, people begin to trust the process and its results, allowing more and more decisions to be made automatically by applying machine learning models.
There is no ‘formula’ for what kinds of decisions should be made based on automated processes or machine learning, versus decisions that need to be made by humans based on data analysis.
After all, machine learning algorithms are probabilistic by nature. There is no guarantee that the desired outcome will happen. When using algorithms to make medical decisions, for example, doctors need to realize they are making “machine assisted” decisions, but that ultimately they have full responsibility. For example, data can be modelled for finding patients with the most likelihood of benefitting from a specific medication. Doctors can then create a shorter prioritized list of customers who might benefit from it based on further analysis. In cases like that machine learning can increase the productivity of the doctor and the probability of success, but of course there is no guarantee the patient will be cured.
Advocate Health Care, for example, is a pioneer when it comes to moving healthcare forward with data mining and analysis. As the largest healthcare provider in Illinois, the company conducts research studies on vast quantities of patient data that inform system-wide, vital healthcare decisions and can clearly demonstrate that there is a positive relationship between the number of procedures performed and quality outcomes.
When it comes to business decisions, machine learning has the greatest impact on situations where many decisions need to be made continuously, and it’s not possible to do so manually. In situations like that machine learning can help increase productivity by prioritizing important points for analysis by humans. In addition, organizations can periodically look at the predictions made by the machine learning algorithms and how they compared with actual results. By constantly evaluating the performance of machine learning algorithms and updating them when needed, organizations can increase their confidence and trust in their analytical capabilities.
Analytics You Can Trust for a Future You Create
SAP has been managing business processes for companies of all sizes and industries for decades, with over 75 percent of the world’s transactions running on an SAP system today. That means vast amounts of customer data reside in SAP systems. Over time, SAP has developed a comprehensive set of solutions for dealing with all forms of data that vary in volume, variety and velocity, and those solutions are continuously evolving to meet the needs of customers and improve people’s lives.
For example, we can’t prevent natural disasters, but we can mitigate their consequences, thanks to technology. By applying machine learning algorithms to earth observation data collected by the European Space Agency, and using the high-performance processing power of SAP HANA in-memory database platform, we can predict disasters that result from natural phenomena. Organizations can then react to natural disasters more quickly and warn local authorities to mitigate further loss and avoid human tragedy.
Additionally, increasing growth in data technologies like IoT, Big Data, and blockchain, solutions like SAP Leonardo allow companies to capture, process and analyze high speed streaming data. Large volumes of different types of data can be combined with transactional data from SAP applications seamlessly. This enables customers to have a consolidated view of their enterprise by aggregating and analyzing information from multiple sources and presenting it in a format which is actionable with drill down capabilities as seen in SAP Digital Boardroom.
At the end of the day, it is up to us humans to determine how we want to use data, machine learning and other technology to improve our lives and create a sustainable future. As Abraham Lincoln and countless others have said, the best way to predict the future is to create it. No matter what business you are in, your future success depends on how well you manage your data and how you well you apply technology that will help you make decisions you can trust.