This.
I'd add that Kaggle is very good for the "other end" of data science: they generally have pretty clean data, and clear problem descriptions.
In real life the data is never clean and the problems are rarely known in advance.
I'd add that Kaggle is very good for the "other end" of data science: they generally have pretty clean data, and clear problem descriptions.
In real life the data is never clean and the problems are rarely known in advance.