Property talk:P2571
Documentation
number of standard deviations (sigma) expressing the confidence level of a value
| Represents | measurement uncertainty (Q1403517) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Data type | Item | |||||||||
| Domain | qualifier to number datatype-properties (note: this should be moved to the property statements) | |||||||||
| Allowed values | 1 sigma/3 sigma etc (note: this should be moved to the property statements) | |||||||||
| Example | C/2020 F8 (SWAN) (Q90914968) → standard deviation (Q159375) tritium (Q54389) → standard deviation (Q159375) 4-pin fan connector (Q65109480) → engineering tolerance (Q950292) | |||||||||
| Lists | ||||||||||
| Proposal discussion | Proposal discussion | |||||||||
| Current uses |
| |||||||||
| Search for values |
List of violations of this constraint: Database reports/Constraint violations/P2571#Scope, SPARQL
List of violations of this constraint: Database reports/Constraint violations/P2571#Entity types
Use this to source error
editWould it be possible to use this property to source/explain an error (other than standard deviation). I was thinking something like uncertainty corresponds to (P2571)significant figure (Q1056761) to xpalin why 1.0 gets the error +/- 0.05.--151.177.40.12 09:19, 31 October 2016 (UTC)
Is this prop sufficient for min and max values for props like size or mass?
edit(Copying my now-archived post from Project chat and one reply to here:)
- Min and max values for properties like size or mass
Can these already be entered somehow? If not, could you please make this possible?
If this is not the right place to ask about or request this, please link where to ask.
Could you add ways to add things like (healthy-/common-/conventional-/standard-/…)minimum and maximum and/or confidence interval 95% value and/or things like that?
As an example and what got me wondering about this is this study which reported the mass of the reportedly likely heaviest animal that ever lived on Earth: Q121095909 (Perucetus colossus). I entered the min and max values they calculated in the study as two separate values in property "mass" but that is not how it should be and was not changed by now nor did I find another way to enter these values.
Moreover, items like Q5 (human) probably should have some structured info on height and mass which could have different types of sourced values such as the 95% CI (with min and max or ±) at some point in time. Humans aren't 20 meters tall and I wondered about this when generative AI often messed up the body sizes of animals. Prototyperspective (talk) 11:49, 20 February 2024 (UTC)
- Interesting topic. The built in quantity type takes an uncertainty interval (example: Q52353), but there are some problems with that, first off while the data model allows the +/- to be unevenly distributed, the user interface won't let you input uneven uncertainty intervals. The other issue is what this interval represents. Some digging revealed there is a uncertainty corresponds to (P2571) qualifier that can be used to specify whether the interval is a standard deviation or a two sigma interval. This should be good enough to input normal distributions at least. There are also upper limit (P5448) and lower limit (P5447), presumably those solve the issue of uneven uncertainty intervals. Infrastruktur (talk) 13:37, 20 February 2024 (UTC)
- @Infrastruktur: Since this post was archived, I copied it here where it can keep getting discussed for longer. If there is another place to ask about it if it doesn't get solved here, please let me know. I think this prop comes closest to what I'm asking about but I don't know if it's sufficient or needs to be altered, complemented by an additional prop, and/or be used more. Questions that remain include:
- example Q54389 (tritium) has only "uncertainty corresponds to[:] standard deviation" set in mass, doesn't it need to also specify which standard deviation like in your example (benzyl alcohol) where it's 2 sigma?
- If the above is an issue lots of other items probably also haven't got this specified so it may be good to scan for this using some query. Additionally it seems not unlikely that this prop is underused and more broadly that even major items do not have mass and size set.
- (You can probably skip this:) For example, I don't know if these should be set on Q146 (domestic cat) but I think they should be and one could specify min and max values and/or the specific species; note that there the "life expectancy" is set with "sourcing circumstances" "circa" Q5727902 – that prop is also related. In general, I'm coming to this from the concept that it's better to have a rough idea of e.g. size and mass than none and that this is useful for example when tasking an AI to draw an image of a cat (with realistic size) even if it could infer the size more inexplicitly from training data images (which btw may contain inaccurate sized cats as in art).
- the mass of Perucetus colossus is still like before; how would it be changed? Please implement this change if already possible. I think the mass they calculated refers to specimens of criteria generally-healthy and adult-fully-grown which maybe could be specified.
- I don't see why uneven uncertainty intervals would need to be entered. For example, if body-size of some animal is unevenly distributed one would enter the mean by the body-sizes that have been measured and specify that and/or the mean of the inferred body-sizes according to body-size associations and correlations models and specify that (it's not important as this is just to get an approximate idea of how large animals roughly are rather than extremely mm-accurate means).
- @Infrastruktur: Since this post was archived, I copied it here where it can keep getting discussed for longer. If there is another place to ask about it if it doesn't get solved here, please let me know. I think this prop comes closest to what I'm asking about but I don't know if it's sufficient or needs to be altered, complemented by an additional prop, and/or be used more. Questions that remain include:
Prototyperspective (talk) 16:58, 27 February 2024 (UTC)
- Any progress on this? This is relevant to mass and size. Relevant study Body size comparisons of fossil species should be based on ontogenetically controlled statistical parameters, rather than simply comparing the largest known individuals whose recovery is highly subject to sampling intensity. I think it would be best to have largest and smallest natural/healthy values for mature adults as well as some best-fit average. Note that diseases can mean there are exceptions above or below these values. Also maybe there should be different values for different life stages. Most important and most-used of all of these would be the average value for some organism (and objects?) because it gives a rough idea and could be used for various applications. Prototyperspective (talk) 16:41, 1 December 2024 (UTC)