Displaying 2 results from an estimated 2 matches for "loesljrg".
Did you mean:
loesberg
2017 Sep 06
0
Interesting behavior of lm() with small, problematic data sets
Indeed (version-specific).
With R 3.4.1 on linux, I get coefficients and residuals that are
numerically exact, F-statistic = NaN, p-value = NA, R-squared = NaN, etc.
All of which is what ought to happen, given that the response variable
(y) is not actually variable.
---JRG
John R. Gleason
On 09/06/2017 09:10 AM, S Ellison wrote:
>> I think what you're seeing is
>>
2017 Sep 06
2
Interesting behavior of lm() with small, problematic data sets
> I think what you're seeing is
> https://en.wikipedia.org/wiki/Loss_of_significance.
Almost.
All the results in the OP's summary are reflections of finite precision in the analytically exact solution, leading to residuals smaller than the double precision limit. The summary is correctly warning that it's all potentially nonsense, and indeed the only things you can trust are