Machine Learning is just Statistics

Machine Learning is just Statistics

This is a very provocative statement 🙂

You will find below the link of an interesting paper from Joe Davison on Towards Data Science about that. I agree that marketers (and I was one of them so I know about it) have always been very creative to generate new buzz words to create new categories and try to lead them. I worked almost 8 years to market Predictive Analytics solutions (@ SPSS and IBM), we could already see at that time the fight between statistician and data miners about what was what.
When we hear about models or algorithms, it sounds about Statistics or about Data Mining (or later called Predictive Analytics).  But Machine Learning as Joe said is different because you are using a model or an algorithms using millions of variables. Is it not just a scale difference? If you are speaking about Neural Networks, is it different in a Machine Learning environment than in Data Mining ? Neural Networks are Neural Networks…Few years ago started the trend of Big Data, it was also confusing. People were speaking or about the IT infrastructure providing the performance needed to compute millions of data in realtime or the application benefiting from the underlying and performant IT infrastructure. But if I want to be as provocative as the post title from Joe… Are we not just speaking about the same thing? Data Mining Models + Big Data Infrastructure = Execution of algorithms at light speed with tons of data… Almost the definition of Machine Learning from Joe 🙂

I am not a statistician nor Data Scientist, so maybe I am missing a lot but from a Marketing view I have the feeling that some of my colleagues created again a nice new sexy funky buzz word for something that « almost » already existed. But for sure we will make billions of dollars selling new Machine Learning applications to companies for their benefit !

Read the Joe’s paper here