Personally, I’m a machine learning enthusiast. I get great satisfaction out of building models that can make good predictions in a particular problem space. However, models are just mappings of inputs to outputs that are derived from data…they’re mathematical representations of the data. So, I’m the first to acknowledge the fundamentally critical role that good data engineering plays in the success or failure of my machine learning endeavors. No amount of model architecture tweaking can ever make up for training data that inadequately represents the problem space. Trying to use data that’s inappropriate for the problem to be modeled, is analogous to pouring old stale gasoline into a rocket engine…you’re not going to get very far.
So, I’m very happy to have found this podcast that casts a proper spotlight on the absolutely critical role of data engineering - the rocket fuel that powers all successful ML/AI endeavors!
