Level of Resources versus Urgency of Problem
The Business of Service Management
Reinventing Electric Power Value Chain
Utility Game-Changers: Solar, Wind, Hydro and Fintech
Will the Smart Meter Deliver on its Promise?
John Burke, CIO, Ambit Energy
IT Governance Built to Last: The Wisconsin Enterprise Model
David Cagigal, CIO, State of Wisconsin
The Role of CIO in the Cloud-First World
Yvonne Wassenaar, CIO, New Relic, Inc
Engaging Citizens through Technology
Martin P.Rose, CIO, Pinellas County
Machine Learning and its Needed Enterprise Framework to Survive
By Sol Rashidi, Chief Data & Cognitive Officer, Royal Caribbean International and Carl Marrelli, Director of AI, Royal Caribbean Cruises
To start, we must begin with a business problem. A formal process that starts with business objectives, design, iterative build and train. Second, we all know a framework with the right balance of data science and engineering is crucial to ensuring any advanced modeling is fueled by the right data at the right time. Therefore data must be captured, ingested, aggregated, and compartmentalized and we need to employ armies of engineers, analysts, architects, ETL experts, and data scientists to do it. Third, since Machine Learning (ML) algorithms can be applied to many facets of an enterprises data, start in the area where you can collect feedback from the outcomes, this process is necessary to continually tune and optimize the models themselves to ensure precision and accuracy over time.
So you’ve got the data, you’ve tuned the model, now what? Fourth, you need to re-engineer your DevOps strategy to include optimizing the ML models as a core function; to make sure scale is possible vertically and horizontally as machine learning permeates the enterprise and “model version and release” becomes as common as “software version and release”.
Data Science as a practice will need to be institutionalized, and the inherent desire to treat this team as advanced report makers needs to be deterred
Last but not least, as important as the technology, is the team in place to help pivot the thinking from data analysis into data science in an effort to scale Machine Learning and its cognitive capabilities. Data science as a practice will need to be institutionalized, and the inherent desire to treat this team as advanced report makers needs to be deterred. This is key for any enterprise to create and curate a data landscape that is conducive for ML applications.
And while we have individual companies who have succeeded in setting up this right framework, our society as a whole is yet to think beyond the ‘buzz’ of A.I. and Machine Learning, and consider this discipline a new normal, a means to surviving the pace of change and the competitive landscape.
As such, it is our duty to educate our executives on the proper framework conducive for Machine Learning survival. And when asked the inevitable questions on, how much will this cost and when can I expect an ROI, the response should not be a quantitative discussion. The first reply back should be which ROI are you referring to because there are three necessary components we need to reflect on, as we advance towards modernization:
• Monetary ROI – revenue potential and savings cost associated with ML applications
• Relevancy ROI – if we don’t do it, we won’t be relevant with the pace of change in our market
• Cultural ROI – pivoting our company from data analysis into data science for market differentiation
As such, while we are all data rich, very few of us are data driven and the applied science of Machine Learning is the start to any of these ROI questions, while the section above describes the right framework.