-
-
Notifications
You must be signed in to change notification settings - Fork 5.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Parameter estimation for : Left skewed distributions #13071
Comments
To my recollection, no. The generic optimization does not check the skew and change the procedure based on that, if that's what you're asking. Should it? I'm not sure I can help with the rest, but I'll try. There is data in a variable
Can you elaborate on what you are expecting and why what you get does not match your expectation? I don't understand what is not making sense. We may not be as familiar with the distribution as you are, so it is not obvious from the numbers what is wrong.
You may be interested in gh-12097. It added an analytical formula for the scale parameter. It will be available when SciPy 1.6 is released, or you are welcome to build SciPy from master now. You might also be interested in this reference; see the fitting information in Chapter 39. The other first-order condition for Rayleigh, which you can use to get the location if you know the scale, is In master, the fit returned by Please also check how |
@digitalgemba I'll go ahead and close this, but if you think there is a bug in master or if you're requesting some other feature, please let me know. |
The data below , is skewed
SkewtestResult(statistic=-3.4412103891299326, pvalue=0.0005791180499256706) The parameters does not reflect actual data . How do we address such issues ? |
I want to make sure I understand your issue. Let me try to restate it: |
Yes. |
Ok, let me try fitting your data. Which distribution do you want me to try? Rayleigh, lognorm, or something else? In your original post you mentioned Rayleigh, but your recent post showed "lognorm" in the plot. |
Please check for lognormal, rest I will try after your feedback. I measure distribution fitting by r^2.
|
But the other way of looking at this might be that lognormal may not be the appropriate distribution. |
I was going through the code for
def fit(self, data, *args, **kwds):
, Does the optimisation differentiate between negative and positive skew distribution ?if yes , please let me know the line link and ignore further.
If not , for a data as below,
temp=([0.035, 0.038, 0.054, 0.049, 0.052, 0.035, 0.035, 0.05 , 0.042, 0.034, 0.052, 0.035, 0.034, 0.037, 0.052, 0.035, 0.041, 0.035, 0.039, 0.04 , 0.04 , 0.043, 0.043, 0.035, 0.037, 0.034, 0.036, 0.037, 0.035, 0.032, 0.034, 0.038, 0.031, 0.05 , 0.042, 0.034, 0.036, 0.037, 0.055, 0.034, 0.036, 0.037, 0.035, 0.055, 0.044,0.04 , 0.031, 0.035, 0.038, 0.054])
For Rayleigh Distribution,
This does not make sense to me as this is case of 0 to loc , rather than 0 to Inf.
I have written the following, and I am struggling with
._fitstart
.Please comment.
The text was updated successfully, but these errors were encountered: