In this article about exact differential equations, for the fourth example, they have the differential equation:
$$ (5xy^2 - 2y) dx + (3xy^2 -x) dy = 0$$
And they say they integrating factor for this is $$ \mu(x,y) = x^a y^b$$
Now, I don't get how they arrived that this should be correct integrating factors for the problem. Of course, this does integrating factor does satisfy the requirement that the second-order partials commute but I can't really understand how I'd come up with this if it was not already told that this is the integrating factor.
If not, for what kinds of differential equations would this integrating factor work?
I have seen this post already and it is not the same as mine because the question was not really regarding the integrating factor which I mentioned but rather alternative methods.