Data Science Lifecycle

Also known as · lifecycle

The iterative loop of Ask, Collect, Clean, Explore, Model, Communicate and Decide, in which each stage can send you back to an earlier one (for example, EDA reveals dirty data, so you clean again).

When to use

Use it to place any task in the pipeline and to justify why most business questions need only EDA, not a model. Up to 80% of a data scientist's time goes on collecting and cleaning, and a vague question guarantees a useless answer.

See Lec 1.

Appears in

No references yet.