In estimation Weibull parameter using parameter prediction or parameter recovery, shape parameter is relatively difficult to model than scale parameter. What is the reason behind it?
Because scale is simply the mean, calculated using simple mathematical functions, but shape describes kurtosis
Weibull modulus, aka the shape parameter, is not difficult to estimate. There are simple techniques to estimate it along with the scale parameter, such as the least squares and method of moments.
I strongly recommend you to use SW such as JMP (for example) for calculation Weibull distribution parameters and it'll take you a couple of minutes to get the distribution parameters
For estimating Weibull parameters you can use:
1) moment method (already suggested by Prof. Tiryakioglu) where you take the log(data) and then calculated moments (the data will be in this way transformed to SEVD);
2) use the ML method: excellent books by Nelson explain it.
It is worth plotting the data onto a Weibull probability paper (once again log(data)): the modulus is the slope of the data.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators