Hello Matthew,

I overall agree with the sentiment of your article. Well written. In fact I have also echoed similar sentiments in my article in the past. https://towardsdatascience.com/so-how-many-ml-models-you-have-not-built-e692f549b163

I do have a small disagreement though.

You have stated “ Machine learning is stochastic, not deterministic”.

I beg to differ on this. Speaking strictly in statistical terms, some machine learning algorithms are not stochastic. Stochastic events are those which can’t be predicted at all. For e.g Brownian motion of gas molecules.

Where as an Machine learning algorithm can make a determination (albeit due to a decision boundary )whether a image is of a cat or dog .

Other examples of stochastic events could be predicting the stock market.

Machine learning by itself is not stochastic, its application to certain problems may not result in success because the problems are stochastic in nature.