Thoughts.

While overfitting is classically seen in data science as a technical phenomanon, that can be solved through proper technique, i.e. split testing/training data, k-fold cross validation, etc... A recent twitter exchange has me thinking about it differently

Galen Caldwell