Liberating to the dogma of "Degrees of Freedom"
Statistics has rudimentarily been about mapping "best" attributes of the population. When modeling the population, it becomes essential that accredition is maintained in terms of the range of values that variegated actors can clasp.
Connotatively, Degrees of Freedom of an expression/ model could be deemed as the number of variables or features that are holding certain quantum of values. Perpetually, the degrees of freedom is a number, one less than the total number of variables in the model.
For exemplification, let us consider a scenario where a candidate is on a mandate to score a minimum of 10 points, across a series of 3 exams to be graded as PASS. This is pretty small-scale, nonetheless gives sense of the idea.
Score(Exam1) + Score(Exam2) + Score(Exam3) ≡ Grade (PASS/ FAIL)
If the student scores 2 in Exam1, 3 in Exam2, she is left with "NO OPTION" but to score atleast 5 in Exam3. Thereby, we can imagine that she has a free hand in the scores of the first two exams, but has to concede to the third one with a definitive constraint. Thus, the model has 3-1, i.e. 2 degrees of freedom.