Emory Solutions

Want to see where you stack up on Data Readiness? Take a look at our data checklist. It's never too late to see how you're doing. https://lnkd.in/geTJiiA4

When data is a liability... https://lnkd.in/gFxPzXQs

Penny Wise and Pound Foolish... Read more.... https://lnkd.in/dYkd-9PD

"The hardest part should be in compliance and non-bias.?I say hard because we should never rest that we have non-bias producing data that will poison our models. The next more difficult item is getting your hands on the right data to begin with.? Models need lots of data and better accuracy requires more features.? So now you have to get datasets that are both trusted and have duration to them."

"Design is more than pretty pictures. When it is intentional and crafted, it creates an Experience, and that?is everything. Every detail matters from the font and font size, colors, arrangement, button placement, and what data is displayed and when. I am always thinking about how to take friction out of the process and create something that delights both internal and external users."

No.?Right now, there’s too many definitions of a data scientist vs. data engineer vs. data analyst.? Different companies have wildly different expectations.? The lack of standardization is affecting the hireability of qualified candidates and resulting in hiring of candidates that aren’t qualified for a particular role. I’ll modify Gartner’s “Trough of Disillusionment” with another peak forming after the hype, which will be a flooded market of people looking to get into the ML game that may be missing the skill.

We recently had a scenario where a regression model was way off.? The data was right, the features were right, but the ML model consistently produced poor scoring.? So instead of leaning that the model was bad, I investigated the data and found a very necessary feature that neither I or the customer would’ve identified on our own. As a result, we found good data for the feature.? The model improved and we identified a new key influencing factor.

"Rapid prototypes and direction.? It clears up a lot of ambiguity and creates stepping stones to know the TCO (total cost of ownership).? Specific to AI/ML, the more realistic you can make those PoC’s the better, so make good passes at getting real data to tell stories they can connect with. Tell that story with good visuals and facts.? It’s a better than fair chance they know their business better than you, so do not torture that data to turn it into some pleasant end.? Finally, come equipped with a plan.? How much data do you need to improve the model?? What kind of support will you need?? Where will the data be housed?? What’s the timeline?"

Challenges and solutions for AI/ML in large organizations.

This is what we live and breathe. https://lnkd.in/dD4PHPnQ #enterprise #software #middleware #development #integration