-
Notifications
You must be signed in to change notification settings - Fork 2.8k
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
iterated lowess is wrong #2108
Comments
@carljv Any comment on this? I'm not familiar enough with this code to say anything without studying it for a bit. |
Well, it's not quite as dire as the title sounds. It could just be in these odd cases, R is doing something different. If you want to dig in here and figure it out, it would be much appreciated. For 'normal' data, we agree with R.
Statsmodels
|
I'm trying to partially figure out how the neighborhood is determined, #2449 it looks to me there is an asymmetry ??? |
However, we are using Also, what's the behavior at the lower boundary. Is x[0] necessarily included in the first interval? |
Does this apply to smoothers_lowess or smoothers_lowess_old? |
It's for the cython code that is actually used. |
So, I looked into the question of indexing. And it looks fine. Let's say you have 30 datapoints and Does this help? |
Consider
(either with the old, 0.6.0 version, or the latest "nan-fixed" version)
This is likely incorrect as the output array isn't even symmetric around 0.5.
Compare with R's result:
which is symmetric.
Note that the issue doesn't appear if
iter=1
(statsmodels gives the same result as R).The text was updated successfully, but these errors were encountered: