A significant portion of ML workloads involve predicting or classifying something. Linear/logistic regression of the right variables/features typically gets a significant portion of the data's ability to predict /classify correctly, while being significantly easier to build, train, deploy, and understand.

Heck, in a large number of domains, simple ratios -- debt to income ratio in finance for example -- will dominate the feature weight for many models and can be used on their own as a pretty good heuristic.