Counting and ML, from Paolo Pezzutti
What is the role of Machine Learning models and Features selection in this "counting" philosphy? Have all these "new" methodologies overcome and made useless the traditional counting and statistical approach? Or can they coexist, as long as one can find a niche in which to conduct profitable operation?
William Huggins responds:
ML and feature selection run on "traditional statistics", which is basically about comparing empirical data to what randomness around a benchmark should look like. think of them as like hydraulics, which transformed the shovel into the backhoe for large operations but without rendering the "basic version" obsolete.
Big Al links:
In the Google Crash Course in Machine Learning, the first model is Linear Regression.