Is It a Rat's Nest?

supervised and semi supervised both are actually harder to do than a completely unsupervised doing. Any time you do labelling, i think that's actually way harder than unsupervised. And theni this line of questioning is coming from another conviction that i teach like all the time: hey, whait? Just so happens that that 80 peret more work we do in the a data pipe line before we get to predictive modelling really gives about 80 % of the value back to the business, if you treat it right.