Timeline for Can I use 'mean ± SD' for non-negative data when SD is higher than mean?
Current License: CC BY-SA 4.0
33 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 16, 2023 at 16:17 | history | edited | OverLordGoldDragon | CC BY-SA 4.0 |
added 3030 characters in body
|
| May 14, 2023 at 10:58 | history | edited | OverLordGoldDragon | CC BY-SA 4.0 |
added 392 characters in body
|
| May 13, 2023 at 13:51 | history | edited | OverLordGoldDragon | CC BY-SA 4.0 |
added 1061 characters in body
|
| May 13, 2023 at 13:22 | comment | added | OverLordGoldDragon |
@mkt It may be on-topic on paper, but I see a different practical reality here. In a community with sufficient familiarity, my answer should be around +5, relative to others. That it was at -3 instead tells me that I have to put in more work than is really needed, like summarizing the Fourier transform to a non-signals audience instead of just applying fft. "If the goal is a measure of spread of data" and "[standard deviation] is a nonlinear alternative to mean absolute deviation, with a quadratic emphasis" pretty much say all of it.
|
|
| May 12, 2023 at 17:13 | comment | added | Nick Cox | @Igor F. That's not a total surprise. I would have thought SD a more common abbreviation in mainstream statistics. No matter. | |
| May 12, 2023 at 15:06 | comment | added | Igor F. | @NickCox "std" stands for the ordinary standard deviation in the Python community: numpy.org/doc/stable/reference/generated/numpy.std.html | |
| May 12, 2023 at 14:23 | comment | added | Nick Cox | (Meta-comment. The comments are themselves repetitive, but unilaterally deleting mine would make some of the others harder to follow.) | |
| May 12, 2023 at 14:21 | comment | added | Nick Cox | "Sometimes STD is best for similar reasons it's normally best." I can't follow this sentence More generally, is STD the name for this two-number summary? | |
| May 12, 2023 at 14:14 | comment | added | mkt | @whuber I'm not claiming that this is a feature engineering question, merely that feature engineering itself is not off topic and that it would require some explanation to connect it more clearly to the question given the audience. I don't think this answer is that far off base, though. IgorF's answer has not attracted downvotes despite its similarity to this one. It makes a clearer and stronger argument and doesn't invoke feature engineering but that answer itself states it was inspired by this one. | |
| May 12, 2023 at 13:53 | comment | added | whuber♦ | @mkt Try as I might -- I have reread the question many times -- I simply cannot see it as a feature engineering question at all. This doesn't look like any kind of cultural clash: the problem is that this answer, however interesting and useful it might be in some other context, simply doesn't belong in this thread. | |
| May 12, 2023 at 9:19 | comment | added | mkt | Feature engineering is very much on topic here (we have a feature-engineering tag with 750 threads), but it tends to come up on more machine learning threads than statistical ones. I think this is a small culture clash happening here over your answer. This thread has attracted a more statistical audience, and your answer probably just needed more explanation to convey the different perspective you are bringing (which I see to some extent in your 'Clarification'). | |
| May 10, 2023 at 19:02 | comment | added | OverLordGoldDragon | @NickCox Hmm I see. I took a look around again. It was my impression that this network exercised the concept, and indeed it does but only implicitly. Feature engineering isn't off topic but I see why reception to my answer was poor; it appears to be about stats, even if I say otherwise. Thanks for your feedback. | |
| May 10, 2023 at 17:48 | comment | added | Nick Cox | Sorry. but I have no idea what "feature engineering" is. Otherwise I can't speak for CV. | |
| May 10, 2023 at 16:57 | comment | added | OverLordGoldDragon | @IgorF. It also has clear weaknesses despite being baked into the definition of some methods, like normalized cross-correlation: sparsity. Inflating rigor standards for a non-conventional answer isn't the way to go. | |
| May 10, 2023 at 16:56 | comment | added | OverLordGoldDragon | @NickCox I've given up worrying about downvotes. Now what I'd like to know is, does this network concern itself with "feature engineering"? I've made it clear that this is a feature, not statistical answer, and that should suffice. If not, I'll refrain from making such contributions. If you've not clicked every link, I know a thing or two about making features - that answer is almost entirely made up. | |
| May 10, 2023 at 9:03 | comment | added | Firebug | @OverLordGoldDragon I didn't downvote, but there is an obvious caveat: if either part of the distribution falling above or below the mean is constant your method will deem the standard deviation there to be zero. Particularly, if it's a bimodal distribution with a marked separation of the modes, your standard deviation will severely underestimate the variability | |
| May 10, 2023 at 7:43 | comment | added | Igor F. | Standard deviation (unsigned) is an old, extensively studied statistical concept with known and very useful statistical properties. In cases where it's not appropriate, one can still use other established measures, like skewness or quartiles. Your asymmetric SD might be useful, but it would probably require a lot of effort to show that it's better than the established measures. | |
| May 9, 2023 at 23:49 | comment | added | Nick Cox | @OverLordGoldDragon You misunderstand me. As said, I have not downvoted this. I have not voted to close or to delete. I am just urging you to improve your answer by making what you are suggesting more clear. You should be worrying about the downvotes, but the downvoters haven't explained their reasons. | |
| May 9, 2023 at 22:57 | comment | added | OverLordGoldDragon | @NickCox On which StackExchange network is "not explained" grounds for deletion? It's a side point that offers potentially useful alternatives. Such a policy wouldn't even be enforceable, "explained" is subjective - I said "for sparse data", to some that's an explanation. | |
| May 9, 2023 at 21:39 | comment | added | Nick Cox | My comment was that sparse means are not explained in this answer, not that the answer is off-topic. Lack of focus doesn't imply irrelevance. | |
| May 9, 2023 at 20:15 | history | edited | Nick Cox | CC BY-SA 4.0 |
edited body
|
| May 9, 2023 at 16:54 | comment | added | user603 | @OverLordGoldDragon: I don't think that's the premise of the question. That's certainly not the tenor of the top voted answer thus far. | |
| May 9, 2023 at 16:43 | comment | added | OverLordGoldDragon | @user603 That goes with any use of STD; the question's premise is that STD is desired. | |
| May 9, 2023 at 16:41 | comment | added | OverLordGoldDragon | @NickCox Unexplained references are off-topic for this site? That's not how a general StackExchange network operates, do you have this site's Meta to back it up? I don't find this productive at all. | |
| May 9, 2023 at 16:40 | comment | added | OverLordGoldDragon | @Firebug I made it up | |
| May 9, 2023 at 16:40 | history | edited | OverLordGoldDragon | CC BY-SA 4.0 |
added 624 characters in body
|
| May 9, 2023 at 15:17 | comment | added | EngrStudent | This shows up in how NIST thinks about Cpk in process capability. itl.nist.gov/div898/handbook/pmc/section1/pmc16.htm | |
| May 9, 2023 at 14:55 | comment | added | Nick Cox | The sparse mean mentioned might or might not be useful and interesting in this thread. I followed the links and failed to find a succinct summary of what it is. Please either expand your explanation here or delete the mention. (I didn't downvote this, but I don't find it obviously focused on the question.) | |
| May 9, 2023 at 14:16 | comment | added | Buttonwood | @user603 Though for a description of asymmetry / tailing of distributions there is skewness, advantageously coupled with kurtosis. | |
| May 9, 2023 at 9:09 | comment | added | user603 | The problem is the mean: departure from symmetry and fair tails renders it insufficient as summary statistic for the whole distribution. | |
| May 9, 2023 at 7:04 | comment | added | Firebug | Where does this "directional standard deviation" come from? | |
| May 8, 2023 at 23:37 | history | edited | OverLordGoldDragon | CC BY-SA 4.0 |
added 84 characters in body
|
| May 8, 2023 at 23:29 | history | answered | OverLordGoldDragon | CC BY-SA 4.0 |
