What the energy industry can learn about Predictive Analytics from the election
I strenuously avoid mixing my LinkedIn world with my Facebook world (and wish others would do the same) but one relevant thought about the election for those who are using or considering Machine Learning-based Predictive Analytics.
Our Founder/CEO, Dr. Noa Ruschin-Rimini, is always reminding us and educating our customers that Machine Learning is science, not magic. A machine can only learn from what's happened, not what has never happened. Now, the fact is that there are amazing correlations in historical data that you might not imagine or be able to identify with traditional statistical methods but expecting Machine Learning models to deliver accurate results without training them with relevant historical data is a recipe for failure. For example, a model won't get energy usage correct for holidays if the training data doesn't include holidays.
This is why it is so important to carefully define the business objectives of a Predictive Analytics system, ask the right questions of it, and provide it with the right data to learn. Then you're on the path to 270! (or whatever your goal is).