The potential business benefits of data analytics are great, but the predictive models can also u201cgo awryu201d and produce unintended consequences if underlying assumptions are faulty or outdated. Companies need a governance strategy to check their analytics applications for flaws. Predictive analytics can speed up business decisions and create a competitive advantage, but the models need better governance to avoid having some predictions turn into business fiascos, according to research supported by the Society for Information Management’s Advanced Practices Council. Data science is a nascent field where organizations are trying to find the best governance approach to mitigate the risks of analytics failures, according to the research paper by academic Michael Goul. The paper is titled “Taming the Wild West of Predictive Analytics and Business Intelligence.” For all the good that data analytics can do, the models can also “go awry, resulting in unintended consequences,” if the underlying assumptions are faulty or become outdated, the report says. Analytics can also produce business results that are controversial or run afoul of legalities, such as anti-discrimination laws. The report summarizes the emerging best practices for governance at various organizations, including Adidas, American Express and State Farm Insurance. “State Farm established a testing culture in which analytic solutions are assessed to determine whether they are performing as designed and promoted,” the report says. “A testing team comprised of a business partner, a statistical designer, an execution specialist, and a structural and creative design expert” run controlled A/B experiments to evaluate the performance of analytics applications. Predictive analytics can speed up business decisions and create a competitive advantage, but the models need better governance to avoid having some predictions turn into business fiascos, according to research supported by the Society for Information Management’s Advanced Practices Council. Data science is a nascent field where organizations are trying to find the best governance approach to mitigate the risks of analytics failures, according to the research paper by academic Michael Goul. The paper is titled “Taming the Wild West of Predictive Analytics and Business Intelligence.” For all the good that data analytics can do, the models can also “go awry, resulting in unintended consequences,” if the underlying assumptions are faulty or become outdated, the report says. Analytics can also produce business results that are controversial or run afoul of legalities, such as anti-discrimination laws. [ Also on CIO.com: 5 interview questions for big data engineers… and how to answer them ] Data analytics models need “accountability, transparency and traceability,” the report says. Governance teams should be set up to assess data quality, security and compliance, for example. One governance approach is periodic “back-testing” of predictive models against new data and assumptions to see if the model still holds up. The report summarizes the emerging best practices for governance at various organizations, including Adidas, American Express and State Farm Insurance. “State Farm established a testing culture in which analytic solutions are assessed to determine whether they are performing as designed and promoted,” the report says. “A testing team comprised of a business partner, a statistical designer, an execution specialist, and a structural and creative design expert” run controlled A/B experiments to evaluate the performance of analytics applications. download Taming the Wild West of Predictive Analytics and Business Intelligence Society for Information Management The Advanced Practices Council, a program for senior IT professionals, fosters independent research on member-chosen topics. Goul is the associate dean for research at the W. P. Carey School of Business at Arizona State University. He is also a professor of information systems, with research interests in business analytics and data model ownership. [ Next read this: QlikView vs Tableau: Which BI tool is best? (enterprise user reviews from IT Central Station) ] Related content feature IT leaders’ AI talent needs hinge on reskilling Most organizations see the need to revamp their training programs to address AI skills shortages — an approach that delivers intangibles hiring can’t provide. By Grant Gross May 31, 2024 7 mins Hiring Generative AI IT Skills feature Skills the Irish Government CIO uses to advance digital transformation In his eight-year tenure as CIO at Ireland’s Department of Public Expenditure and Reform, Barry Lowry always had a vision of what digital government could look like. Here, he details how an approach built on transparency and innovation is conti By Ian Campbell May 31, 2024 8 mins CIO Government IT Cloud Management brandpost Sponsored by Cisco 3 reasons you should adopt cloud monitoring Cloud network management offers increased security, operational efficiencies, and more. By D Matthew Landry May 30, 2024 4 mins Machine Learning opinion Faultless with serverless: Cloud best practices for optimized returns What does a well-defined serverless approach look like? Let's learn some of the best modern approaches to handling Enterprises and SMEs growing serverless computing needs. By Yash Mehta May 30, 2024 5 mins Serverless Computing PODCASTS VIDEOS RESOURCES EVENTS SUBSCRIBE TO OUR NEWSLETTER From our editors straight to your inbox Get started by entering your email address below. Please enter a valid email address Subscribe