It becomes a line because when you put an exponential function inside its inverse (logarithms) function, the functions sort of cancels out each other and just leaves a simple linear function y = x f(x) = a^x f^(-1)(x) = log(x) y = f^(-1)(f(x)) y = x
Logarithmic scaling is one of the most important thing in an engineer's everyday life. However, from the title I was expecting a way of fitting data that is not linear in a linear, logarithmic or double-logarithmic plot. A general transformation if you like. I'm sure that's possible.
It becomes a line because when you put an exponential function inside its inverse (logarithms) function, the functions sort of cancels out each other and just leaves a simple linear function y = x
f(x) = a^x
f^(-1)(x) = log(x)
y = f^(-1)(f(x))
y = x
Logarithmic scaling is one of the most important thing in an engineer's everyday life. However, from the title I was expecting a way of fitting data that is not linear in a linear, logarithmic or double-logarithmic plot. A general transformation if you like. I'm sure that's possible.
why we must transforming y to log(y)?
So like
Linearization??
Sal, your a great person.
How to un-wind it?
amazing!
I love doing log(y)
Great work!!!
😍😍😍😍😍😍😍😍😍😍😍😍😍😍😍, Just beautifully said
could you instead exponentiate x?
sure you can you just need to find the easiest transformation even if it requires squaring x or exponentiate y
In which playlist-course this video belongs?
www.khanacademy.org/math/ap-statistics/inference-slope-linear-regression/transformations-for-linearity/v/transforming-nonlinear-data
why does it happen