The goal of the experimentation phase is to develop models that
solve you data science use-cases. Model development is done in 4 step cycles:

  1. (Re)define scientific goal
  2. Collect, inspect and prepare data
  3. Create and test model
  4. Validate and discuss results

In the Exploration phase use cases have already been defined on a business level.
At this stage we define scientific goals for each case. The scientific goal has a
clear and quantified target by which we can measure if the data science
model’s results are good enough to produce business value.

When the data science model produces valuable results, it is ready to productize.