How to simplify modeling? After decades of work building models with AI tools, what have we learned? Over the years, there have been many reports that very simple models can perform exceptionally well. Yet, where are the researchers asking “say, does that mean that we could make modeling simpler and more comprehensible?”

To fill that gap, we offer EZR, a simple, fast toolkit that can model complex problems with just a few data points (using incremental sampling). The approach supports classification, regression, optimization, fairness, explanation, data synthesis, privacy, compression, planning, monitoring, and runtime certification (but not generative tasks).

For all these tasks, our minimal data usage simplifies model construction and verification. The lesson from all this work is not everything can be simplified, but many things can. When simplicity works, we should embrace it. Who can argue against that?

We illustrate EZR with 49 examples from the SE domain ranging from (a)~the control of software processes to (b)~controlling the models learned by AI tools to (c)~the configuration of video encoders. With EZR, certain tasks that were previously intimidating complex (e.g. hyperparameter optimization) are now practical, easy, and fast.

All the materials used here are available on-line under and open-source license (BSD2). This content would be suitable for a one semester advanced SE graduate class on ``how to refactor AI to make it simpler''.

Full prof, ex-nurse,rocketman,taxi-driver,journalist (it all made sense at the time).