iterative.

linear.

it communicates the outputs from it in such a way that colleagues with a nontechnical background can understand them and have confidence in them.

it justifies the extensive investment required for an ML project.

it helps to hide the uncertainties surrounding analytic results, if the data is not reliable.

must be removed from the data set(s) before any analytics operation.

is not of interest to analytics because it is not the captured data that holds the core meaning.

can be used to reveal hidden patterns and insight into situations.

we want to automate behavior patterns or operative processes and we do not yet know how to do that manually.

we face a problem that humans cannot solve, but we have access to data sources that might be of help.

there is the need to safe costs related to the use of information systems.

we have a large number of data examples and when the patterns are too complex for humans to discover and extract manually.

a real-world, domain-specific problem and the need to design a data-driven solution to this problem.

a hard calculatory challenge that we cannot solve using statistical means but for which sufficient data is available.

a new programming technique that we can test on interesting practice problems.