As the header suggests, calculations that use Boolean values or numbers are more performant than those that use dates. Calculations that use dates, in turn, are more performant than those using strings. This is not only true of Tableau but also in computer science as a whole.
Based on this information, Scenario II listed in the table above is more performant than Scenario I. Scenario I causes Tableau to create a query that requires the data source engine to handle strings for reporting profitability, whereas Scenario II sends only 1s and 0s to determine profitability. The third step for Scenario II (that is, aliasing 'True' and 'False' to 'Profitable' and 'Unprofitable') is merely a labelling change that happens after the aggregate dataset is returned from the data source. Labelling is quick and easy...