Timeline for Variance estimate for Student's t-distribution with heavy tails
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 14, 2022 at 14:54 | history | edited | kjetil b halvorsen♦ | CC BY-SA 4.0 |
added 64 characters in body
|
| Apr 11, 2022 at 20:38 | history | edited | jbowman | CC BY-SA 4.0 |
added 125 characters in body
|
| Apr 11, 2022 at 16:44 | comment | added | Sextus Empiricus | If you know that the distribution follows some parametric form, for instance that it must be a t-distribution, then you can use this to estimate parameters and the variance. However, if the distribution is just heavy-tailed and if you do not necessarily know which type of distribution the data is sampled from, then those robust estimators will not help a lot. | |
| S Apr 11, 2022 at 12:00 | history | suggested | Soltius | CC BY-SA 4.0 |
direct link to the answer
|
| Apr 11, 2022 at 11:47 | comment | added | whuber♦ | A suitable multiple of the IQR would be good if you know the df. If you don't know the DF, then you will need a robust estimator of it and proceed from there. There are many more sophisticated approaches available, too, but using an IQR (or any inter-percentile range) would be exceptionally simple and likely work pretty well. | |
| Apr 11, 2022 at 9:46 | comment | added | COOLSerdash | What would be a good robust estimator of the standard deviation in this case? | |
| Apr 11, 2022 at 6:49 | review | Suggested edits | |||
| S Apr 11, 2022 at 12:00 | |||||
| Apr 10, 2022 at 21:41 | vote | accept | sayda | ||
| Apr 10, 2022 at 20:42 | history | answered | jbowman | CC BY-SA 4.0 |