Timeline for Fitting and plotting Weibull distribution
Current License: CC BY-SA 4.0
22 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 29, 2022 at 19:55 | history | edited | Carl | CC BY-SA 4.0 |
added 196 characters in body
|
| Jul 29, 2022 at 19:30 | history | edited | Carl | CC BY-SA 4.0 |
added 261 characters in body
|
| Jul 12, 2022 at 7:09 | history | edited | Carl | CC BY-SA 4.0 |
added 183 characters in body
|
| Jul 12, 2022 at 5:56 | history | edited | Carl | CC BY-SA 4.0 |
added 814 characters in body
|
| Jul 12, 2022 at 5:49 | history | edited | Carl | CC BY-SA 4.0 |
added 814 characters in body
|
| Jul 3, 2022 at 6:39 | history | edited | Carl | CC BY-SA 4.0 |
added 32 characters in body
|
| Jul 2, 2022 at 8:44 | history | edited | Carl | CC BY-SA 4.0 |
added 1962 characters in body
|
| Jun 30, 2022 at 22:46 | history | edited | Carl | CC BY-SA 4.0 |
added 515 characters in body
|
| Jun 30, 2022 at 21:08 | comment | added | Carl | @JimB I acknowledge that what I do is not what many people believe, I do not base my decisions upon belief but am convinced by primary evidence. | |
| Jun 30, 2022 at 21:05 | comment | added | Carl | @JimB Yes, I know you think that. I am worried about misuse of procedure just like you are, it's just that my experiences, and the testing I've done did not lead me to the same place as it did for you. You said you were going to try the Poisson norm, please do. I am willing to investigate your claims to exhaustion, do the same for me please. | |
| Jun 30, 2022 at 18:27 | comment | added | JimB | I think you're throwing in roadblocks that just are unnecessary for this particular question. There are only 2 or 3 parameters here. But because regression seems to be a go-to procedure, the best book/advice out there is google.com/books/edition/Regression_Modeling_Strategies/… and there's a video from the author at youtube.com/watch?v=CwGyoo-D8iY. | |
| Jun 30, 2022 at 17:02 | comment | added | Carl | @JimB There are lots of measures of goodness-of-fit, and, they all have warts including AIC. No, I don't rely on AIC, I have lots of posts that show why that is. For instance, AIC is only asymptotically valid, and unless you have hundreds of data points, and very few parameters, you may be asking for trouble. More, R$^2$, adjusted R$^2$ BIC and the lot of them break down when the number of parameters approaches the sample size. The best way I have found to compare models is to compare the answers to against general functional models of the same number of parameters. | |
| Jun 30, 2022 at 16:33 | comment | added | JimB | One approach to obtain a relative measure of goodness-of-fit is to use AIC (or AICc). But to use that measure, you'll need to use maximum likelihood. See en.wikipedia.org/wiki/Akaike_information_criterion. | |
| Jun 30, 2022 at 15:42 | history | edited | Carl | CC BY-SA 4.0 |
added 41 characters in body
|
| Jun 30, 2022 at 15:34 | history | edited | Carl | CC BY-SA 4.0 |
added 715 characters in body
|
| Jun 30, 2022 at 15:25 | history | edited | Carl | CC BY-SA 4.0 |
added 715 characters in body
|
| Jun 26, 2022 at 23:08 | comment | added | JimB | I've set up a chat room as you have suggested: chat.stackexchange.com/rooms/137349/room-for-jimb-and-carl | |
| Jun 26, 2022 at 17:03 | comment | added | JimB | However, I will concede that your approach can get one starting values for an appropriate approach such as maximum likelihood or the method of moments. | |
| Jun 26, 2022 at 17:01 | comment | added | JimB |
Let me count the ways: (1) the sample size is (almost certainly) 556 (the total frequency count) and not the number of bins (50), (2) The regression you use assumes a constant variance which is certainly not true (relative frequencies close to 0.5 have higher variance although somewhat reduced by having more observations), (3) Because of (1) and (2) the estimates of precision will not be appropriate – although FindFit doesn’t produce any estimates of precision which is another reason for not using FindFit. I'm sure I can find more reasons.
|
|
| Jun 26, 2022 at 6:49 | comment | added | Carl | @JimB Regression is also not inappropriate for several reasons. For one thing, it is likely MVUE given equidistant $x$-axis data as contrasted to a random variate. That is, there is no $x$-axis random variate, only $y$-axis variability such that least squares in $y$ is not inappropriate. I would like to see why you think this inappropriate, for example, MLR would assume random variate structure of the $x$-axis data, which information was destroyed by binning. So, please explain your reasoning, and if you wish we can continue this elsewhere. | |
| Jun 26, 2022 at 6:25 | comment | added | JimB | This is not a regression problem and ‘FindFit’ is inappropriate for several reasons. | |
| Jun 26, 2022 at 5:28 | history | answered | Carl | CC BY-SA 4.0 |