🧵 Untitled Thread
Anonymous at Mon, 16 Sep 2024 16:58:40 UTC No. 16383474
I was doing a Shapiro Wilk test on temperature measurements for a certain week. The null couldnt be rejected. Then I used distfit library to identify the possible distribution the data comes from. the highest score (lowest residual sum of squares) was for the general extreme distribution. So the data is estimated to come from the genextreme distribution with distfit.
what should I do now? Do I trust the shapiro test or the results from distfit? I looked online for year temperature readings and visualization suggests that temperatures are normally distributed. but also, the general extreme distribution looks like a skewed normal distribution. is city temperature normally distributed?
Anonymous at Mon, 16 Sep 2024 17:00:07 UTC No. 16383478
>>16383474
i forgot to mention that I only have 14 temperature measurements for that week (7 for daytime and 7 for nighttime). These were obtained from the National Weather Service
Anonymous at Mon, 16 Sep 2024 17:35:12 UTC No. 16383524
>>16383474
ignore the shapiro wilk test
Anonymous at Mon, 16 Sep 2024 17:41:41 UTC No. 16383531
>>16383474
You must be new to data analysis to think something is normally distributed.
Anonymous at Tue, 17 Sep 2024 12:43:29 UTC No. 16384576
>>16383474
>visualization suggests
theres your issue mate, either be happy with eyeballing or stop crying about formulas giving you unintuitive results, these 2 approaches are usually exclusive in statistics