“We know from our chief data officer research and surveys that data and analytics leaders that are involved in strategy development are able to deliver far more business value than those who aren’t.”

“Data and analytics leaders struggle to optimize business value, and there’s often a lack of alignment to mission-critical priorities.”

“You as data and analytics leaders need to function as your organization’s ‘chief value officers.’”

“The ROAR (Risk, Opportunity, Appetite, and Return) Model gives you a systematic way to score both benefits and risks for each initiative and then think about the combination of those initiatives in terms of an optimal portfolio.”

“While we have models that can help you think through and prioritize, it’s what goes behind those models that’s important. Thinking trumps doing.”

“Optimizing value from data and analytics is a continuous process. It’s not just a once a year activity, it’s not just a quarterly activity — you need to think about it as central to everything else that you manage.”

Learn how to use data & analytics to re-engineer decision making in the free Gartner e-book “The Future of Decisions.”

“Many projects fail to make it to production due to insufficient or conflicted real-world training data. Organizations can use a variety of techniques to tackle the challenge of low availability of training data.” 

Technique #1 - Expand your data collection strategy. The data which is currently being ignored or dark data can be the missing piece of what your machine learning application needs to know. 

Technique #2 - Spend more time on data preprocessing. For example, removing outliers, finding missing variables, oversampling, etc. Along with these, many data science platforms contain tools and features to support this preprocessing. 

Technique #3 - Data sharing. Create national or international alliances that collect data together and use federated learning approaches so that data can be shared amongst the enterprises without compromising the data privacy. 

Technique #4 - Crowdsourcing. Cautiously crowdsource data not from professional data providers, but from a pool of internet users. 

Myth #1: AI capabilities surpass human capabilities. “The deception is that they allowed the computer top five accuracy metrics. This means that the computer was allowed to make five guesses, and only one of those guesses has to be correct for the system to score ‘correct.’”

Myth #2: AI is disrupting industries. “When you apply AI it has so many use cases, which all will result in better cost savings, customer satisfaction, downtime reduction, risk reduction and more, but all these things that we’re going to see be better don’t necessarily translate into huge disruption.”

Myth #3: AI is about intelligence. “The systems that we create don’t understand much, they only react.”

Myth #4: AI can do anything. “The spectrum of problems that AI can tackle is surprising, but we have to say that AI is a point solution. A solution for fraud detection is not going to drive cars.”

Myth #5: AI will replace human intelligence. “With our flexibility to learn things, to understand things, and to be super fast and adaptive, it will take a long while before AI will replace human intelligence.”