In regression, do i need uniform distribution of output?
I’m working on above example and there is some saying i don’t understand. First, in “Check Data Normalization”, there is saying that “the response (the rotation angle in degrees) is approximately uniformly distributed between -45 and 45, which works well without needing normalization”. I understand why i need normalization but i don’t understand unform distribution of output(=response) works weel without needing normalization.
And second, there is also saying that “If the distribution of the input or response is very uneven or skewed, you can also perform nonlinear transformations (for example, taking logarithms) to the data before training the network”. Here, i don’t understand why uneven and skewed distribution are not going to work well.
NOTE:-
Matlabsolutions.com provide latest MatLab Homework Help,MatLab Assignment Help for students, engineers and researchers in Multiple Branches like ECE, EEE, CSE, Mechanical, Civil with 100% output.Matlab Code for B.E, B.Tech,M.E,M.Tech, Ph.D. Scholars with 100% privacy guaranteed. Get MATLAB projects with source code for your learning and research.
>> I understand why i need normalization but i don’t understand unform distribution of output(=response) works weel without needing normalization.
This is not a general statement. It just happens to work for some problems when parameters are neither tiny nor huge.
There are always tradeoffs.
1. Uniform sampling simplifies the math. However some times the mesh size has to be too small to characterize all of the important info. In those cases sometimes nonuniform sampling is the better choice.
2. Mathematical accuracy tends to improve when parameters are restricted to certain intervals. Whether or not normalization is useful depends more on the extent and variation of the original measurements and parameters.
SEE COMPLETE ANSWER CLICK THE LINK
