I'm a self-taught student in mathematics, and I only began self-study about half a year ago. I was thinking about the harmonic mean, and the logic behind its use. Firstly, let's begin with the arithmetic mean which equals the sum of the set of numbers divided by the number of numbers. For example: say you have a set $(4, 3, 2)$, and you want the mean. You simply divide: $\frac{4+3+2}{3}$. You get 3 in this case. Thus, in a way, the logic behind is to generalize it. It's to take all of the numbers, and distribute them equally to these 3 "slots" if that make sense. In doing this, you "generalize" all the numbers to one single number if that makes sense.
The geometric mean is easy to understand too. Imagine you have a set of numbers which are constantly being multiplied. Say you have a set whose first number is multiplied by 1.3, 1.2, and 1.5. So, starting with 4: $(4, 8, 12, 36)$. We can "generalize" this as we did with the arithmetic mean, because: $4*8*12*36=x^4$. So, we can "generalize" those numbers to the variable x like we did with the arithmetic mean. In the words of a Stackoverflow user by the name of Domagoj Pandzha (who I thank for allowing me to rationalize the geometric mean):
"We are, essentially looking for a value which, when multiplied by itself 3 times (the number of percentages per respective years), gives the same final value as if we simply repeatedly multiplied each year with the respective 1.3, 1.4 and 1.5."
So, $\sqrt[4]{x^4}$ = $\sqrt[4]{4*8*12*36}$ Makes sense to me to use it, since we're trying to average out the products here.
But the harmonic mean I struggle with in my mind. What is the logic behind its use?
Thanks.