I'll play devils advocate and disagree. I'll claim that deep learning is the greatest hammer I have seen in my lifetime and it can tackle all nails you throw a

I'll play devils advocate and disagree. I'll claim that deep learning is the greatest hammer I have seen in my lifetime and it can tackle all nails you throw at it (universal approximation theorem). Additionally it is easy to apply and get very reasonable results. At least for me it's easier to whip up a Jupyter Notebook and use some basic heuristics and a library like fastai than to struggle through Excel. Assuming that the data is available and curated (I'd argue this is a hard and very value creating task).