Re: [OSM-talk-be] Address stats in Belgium
If you use the tool, you know that fixAddresses plugin (who follows what the wiki told them) takes care of the postal code in one simple click on the 'guess button, the editor just needs to verify them. The postalcode + street are the most important ones to put on a building. On Tue, Apr 16, 2013 at 6:46 AM, Jo winfi...@gmail.com mailto:winfi...@gmail.com wrote: I thought the consensus was that we repeat addr:street for each house but not the other information like addr:city, addr:postcode, addr:country and whatnot. Imagine what happens to the size of the DB (and all its derivatives like the planet files), if everybody starts doing that for each and every house/address in the world! That way you will end up with wrong postal codes for several streets, after all -again-: it's not repeating, it's detailing the data. I think you better stop worrying about the size of the data, if people put in stuff like individual tree's and firehose's, parking meters, amenity=bench using 2 nodes... I think it's ok to put in way more interesting data like address data. It's not like it's an extra node, it's a tag. I think the most wet dream for the openstreetmaps founders could have in 2005 is having address data for every building in the world. That is exactly what we are going for! Concerning the consensus, I can't find any stating we should limit what we enter due to data size concerns. It's already huge as it is, but so are cheap disks these days. Potlatch has its limitations, but by choosing to work with it, people indicate a willingness to live with those limitations. I can only hope the IDeditor will overcome those limitations one day and that all Potlatch users will migrate towards it when it does. Jo +1 We should not map for the renderer, nor for the (simpliest) editor. +1 +1 also on not repeating the same data over and over. -1 We should also not map what we think is more convenient for us. We should not map for the machine, but for the 'users', a broad term. Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
On 04/16/2013 05:41 AM, Marc Gemis wrote: I use most of these plugins. But recently I started using them less, because now I convert my GPX waypoints to OSM data points automatically. The only thing I do not do is repeating the postal code over and over, but since you insist, I'll do that from now on. :-) Can you look at e.g. http://www.openstreetmap.org/?lat=51.16279363632202lon=4.425258636474609zoom=16 an area I mapped this winter. Please let me know if you think it can be improved. Yes, the city and the postal code are only in the relation. Some very nice work there, pretty detailed. Looks good on the map. That has been a lot of work by the looks of it. I'm impressed haven't checked with josm yet, but I will. I am going to dig deeper into the Nominatim scene concerning geocoding (=what is best for both map and other data use) , I'll come back on this. For me, the plugin's make me do it, since it's easy. I didn't do all the detailed work before knowing them. I would not recommend doing this manually per building without the things I mentioned (The mapcss helps the most). Yes, I know I should use building=house more consistently. Yes, I know I could add sidewalk, lit, parking lane tags as I did (already partially) in http://www.openstreetmap.org/?lat=51.134501695632935lon=4.385626316070557zoom=16 You mean building=yes (I don't use house, don't think it was 'official'). What I do for sure is mark building=garages and sheds and I remove all addr:* tags from them, as they clutter searching for an address. They make the data worse, and then others think it's an unnumbered building and start inventing numbers (seen that here!) , while all that lives in there are cars or lawnmowers. Let me know if you think the data can be improved, as I'm willing to improve my tagging habits. Cool! The only suggestion I can make is have an extra window in the browser open on the AGIV site, I found out I put my tags on the wrong side in a street using it. A very small one, I had the odd/even sides all wrong. There you can verify if what you enter makes sense, you just can't copy it over without some live visit -ever- as they are incomplete and wrong sometimes, and also don't always know about sub addresses (100b 110/1 etc). Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
I never put address details on sheds or garages. The building=house is on the http://wiki.openstreetmap.org/wiki/Map_Features#Building page. That's as official as it can get for me. Also use building=apartment whenever I can/remember/wrote down. I did use AGIV for some of my more recent expeditions. Unfortunately, it did not help me in a few cases. I had numbers from mailboxes on the street, but didn't know the houses (in private area). AGIV had none of them. With my new workflow (address nodes generated from GPX waypoints), I first have them in a separate layer. I use the lasso tool to select all nodes in 1 street. Then add street (but could easily add city, country, postcode) as well to the whole selection. No need for any of the plugins. But I used them before (for the work in Aartselaar e.g.). A question regarding houses without numbers (e.g. churches, libraries, ...) The official address is e.g. Kerkstraat z/n How is that mapped ? addr:housenumber = z/n does not sound correct to me. m. On Tue, Apr 16, 2013 at 9:32 AM, Glenn Plas gl...@byte-consult.be wrote: On 04/16/2013 05:41 AM, Marc Gemis wrote: I use most of these plugins. But recently I started using them less, because now I convert my GPX waypoints to OSM data points automatically. The only thing I do not do is repeating the postal code over and over, but since you insist, I'll do that from now on. :-) Can you look at e.g. http://www.openstreetmap.org/?** lat=51.16279363632202lon=4.**425258636474609zoom=16http://www.openstreetmap.org/?lat=51.16279363632202lon=4.425258636474609zoom=16an area I mapped this winter. Please let me know if you think it can be improved. Yes, the city and the postal code are only in the relation. Some very nice work there, pretty detailed. Looks good on the map. That has been a lot of work by the looks of it. I'm impressed haven't checked with josm yet, but I will. I am going to dig deeper into the Nominatim scene concerning geocoding (=what is best for both map and other data use) , I'll come back on this. For me, the plugin's make me do it, since it's easy. I didn't do all the detailed work before knowing them. I would not recommend doing this manually per building without the things I mentioned (The mapcss helps the most). Yes, I know I should use building=house more consistently. Yes, I know I could add sidewalk, lit, parking lane tags as I did (already partially) in http://www.openstreetmap.org/?**lat=51.134501695632935lon=4.** 385626316070557zoom=16http://www.openstreetmap.org/?lat=51.134501695632935lon=4.385626316070557zoom=16 You mean building=yes (I don't use house, don't think it was 'official'). What I do for sure is mark building=garages and sheds and I remove all addr:* tags from them, as they clutter searching for an address. They make the data worse, and then others think it's an unnumbered building and start inventing numbers (seen that here!) , while all that lives in there are cars or lawnmowers. Let me know if you think the data can be improved, as I'm willing to improve my tagging habits. Cool! The only suggestion I can make is have an extra window in the browser open on the AGIV site, I found out I put my tags on the wrong side in a street using it. A very small one, I had the odd/even sides all wrong. There you can verify if what you enter makes sense, you just can't copy it over without some live visit -ever- as they are incomplete and wrong sometimes, and also don't always know about sub addresses (100b 110/1 etc). Glenn __**_ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.**org/listinfo/talk-behttp://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
On 04/16/2013 10:08 AM, Marc Gemis wrote: I never put address details on sheds or garages. Didn't mean to sound like you did. I actually did in the past by not paying enough attention to the plugins, I'm in the process of fixing this btw :) The building=house is on the http://wiki.openstreetmap.org/wiki/Map_Features#Building page. That's as official as it can get for me. Sorry, I had to check before talking. The building key is open for free tagging I think anyway, so it would not be an error anyway if it's not on there (yet). Also use building=apartment whenever I can/remember/wrote down. nice suggestion, I'm going to start doing this too. I did use AGIV for some of my more recent expeditions. Unfortunately, it did not help me in a few cases. I had numbers from mailboxes on the street, but didn't know the houses (in private area). AGIV had none of them. you probably know this , I noticed only the deepest zoom level has all the numbers they know about. But I second that it's not the address bible I first thought it would be. With my new workflow (address nodes generated from GPX waypoints), I first have them in a separate layer. I use the lasso tool to select all nodes in 1 street. Then add street (but could easily add city, country, postcode) as well to the whole selection. I pretty much do the same using nodes. But then I use the terracer plugin, it's not only awesome to split houses in 1-2-n pieces but you just select a street + an addr node and a building and press SHIFT-T. It will put the street name in the the building tags. Next I use fixaddresses to complete the other tags, some logic is used to auto-guess the streets, which sometimes fails miserably but all the rest is filled out for you and most of it is correct, it just needs a human to apply the guesses/changes. No need for any of the plugins. But I used them before (for the work in Aartselaar e.g.). Not convinced yet I see :) A question regarding houses without numbers (e.g. churches, libraries, ...) The official address is e.g. Kerkstraat z/n How is that mapped ? addr:housenumber = z/n does not sound correct to me. I would omit the housenumber, that's what I do. tagging it as 'z/n' is the same as putting an imaginary number on it, e.g. it's an error. What I find difficult is buildings with an address containing a business (shop/amenity) with a separate address from the building (or the same sometimes, but the building itself has several numbers). There I see the merit of not duplicating data... I eventually settled for tagging the adress information on the building (less likely do disapear) and putting the amenity within the building border (and try to remove existing duplicate address data on that node). Businesses are far more likely to be gone in 10 years than buildings, when you put the address data on the node instead of the building, chances are great that the address info will be deleted too in the process. But putting the same address info on both (I tried some localy - it's ugly and I need it fixed soon) is very ugly on the map (duplicate housenumbers visible).It makes sense for me, when the shop is gone, you can remove the node entirely, and all address info related to the building is still there. Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
On 04/16/2013 11:01 AM, Marc Gemis wrote: Maybe I should start using the fixaddresses plugin I've been using the plugins terracer, building tool, etc. for over 1 1/2 year. I even wrote a page on it: http://wiki.openstreetmap.org/wiki/User:Escada/JOSM_and_Housenumbers :-) But now that I have a python script to convert my GPX trails, they do not fit perfectly in my workflow. I start from nodes that are tagged with building=house and addr:housenumber=XXX (after importing my modified GPX trail). I still use the terrace plugin to split though. I do use the stylesheets too. If the nodes area already a building node, I understand why that messes things up. you will probably loose information when using terracer on them (there is some precedence). What I really like about it is that when selecting the trio (ctrl-select) street/addrnode/building(way) , after pressing SHIFT-T, it keeps the street selected, that saves you a click when doing the next building. Did you try looking up businesses represented as nodes in buildings with address information ? Nomatim does not understand this (I think). Should we map for Nomatim, not sure ... I just did this:century 21 compas , nominatim finds it right back but you're right from the results set it looks like it didn't pick up the housenumber of the building that encompasses it. bummer, as that could count for a implicit relation between both. There are still some duplicates in there on my fix list, like this take-away chinese restaurant: Xing Fu Lou There both building + node have housenumber, which I think I should remove on 1, but that was using the logic we just debunked. The other way around doesn't seem to function either : Brusselsesteenweg 86, Zemst, finds back the house but not the amenity/shop. The question now is, how can we 'fix' this? Duplicate address information will fix nominatim, but it messes up the map. We should defenitely not map for nominatim more than we map for the map. The difference is that nominatim is a lot more sensitive to the way data is presented/available than the map is. The latter is slightly more 'forgiving' as it's a visual thing, more than a content thing. So by nature it's more sensitive to this sort of situations. Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
I ended up adding all the shop tags on the building. It only works for buildings with only 1 shop/restaurant/pub/... m. On Tue, Apr 16, 2013 at 11:25 AM, Glenn Plas gl...@byte-consult.be wrote: On 04/16/2013 11:01 AM, Marc Gemis wrote: Maybe I should start using the fixaddresses plugin I've been using the plugins terracer, building tool, etc. for over 1 1/2 year. I even wrote a page on it: http://wiki.openstreetmap.org/** wiki/User:Escada/JOSM_and_**Housenumbershttp://wiki.openstreetmap.org/wiki/User:Escada/JOSM_and_Housenumbers :-) But now that I have a python script to convert my GPX trails, they do not fit perfectly in my workflow. I start from nodes that are tagged with building=house and addr:housenumber=XXX (after importing my modified GPX trail). I still use the terrace plugin to split though. I do use the stylesheets too. If the nodes area already a building node, I understand why that messes things up. you will probably loose information when using terracer on them (there is some precedence). What I really like about it is that when selecting the trio (ctrl-select) street/addrnode/building(way) , after pressing SHIFT-T, it keeps the street selected, that saves you a click when doing the next building. Did you try looking up businesses represented as nodes in buildings with address information ? Nomatim does not understand this (I think). Should we map for Nomatim, not sure ... I just did this:century 21 compas , nominatim finds it right back but you're right from the results set it looks like it didn't pick up the housenumber of the building that encompasses it. bummer, as that could count for a implicit relation between both. There are still some duplicates in there on my fix list, like this take-away chinese restaurant: Xing Fu Lou There both building + node have housenumber, which I think I should remove on 1, but that was using the logic we just debunked. The other way around doesn't seem to function either : Brusselsesteenweg 86, Zemst, finds back the house but not the amenity/shop. The question now is, how can we 'fix' this? Duplicate address information will fix nominatim, but it messes up the map. We should defenitely not map for nominatim more than we map for the map. The difference is that nominatim is a lot more sensitive to the way data is presented/available than the map is. The latter is slightly more 'forgiving' as it's a visual thing, more than a content thing. So by nature it's more sensitive to this sort of situations. Glenn __**_ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.**org/listinfo/talk-behttp://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
I ended up adding all the shop tags on the building. It only works for buildings with only 1 shop/restaurant/pub/... Doesn't that remove the nice icon on the map when done on a way ? I really like those icons. Just found an example: http://www.openstreetmap.org/?lat=51.027472lon=4.478946zoom=18layers=M standaard boekhandel ijzerenleen is on nr 55. No icon there, correctly tagged. There is an icon in JOSM, but not on the map ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
Your complain about street being placed in wrong cities, is exactly why we should use associatedStreet relations instead of repeating addr:street addr:city over and over on individual buildings. In that case you only have to correct it once, on the relation, and the data is corrected. But I'll admit that I've been to lazy to add associatedStreets recently, as nobody seems to care. And the tools support outside JOSM is not that great. Repeating the street name twice is completely useless IMHO. I think it is possible to combine lot's of house numbers and accuracy, at least I hope that's what I leave behind :-) m On Mon, Apr 15, 2013 at 5:48 PM, Glenn Plas gl...@byte-consult.be wrote: On 04/15/2013 02:38 PM, JorenDC wrote: Hi, In December there was a thread (start: http://lists.openstreetmap.org/pipermail/talk-be/2012-December/003367.html) containing some numbers/stats. @Sander, *: is it possible to share your used method to pull these stats (or just pull them again) and publish them on a 'frequent' base (I'm not saying weekly, but what about +/- every 4 moths or so)? In that way we can see a bit of our progress regarding this 'project'. The method (overpass query) is mentioned in that link thread (Jo points it out). On the subject. I've been mapping a lot of housenumbers lately, verifying my data against AGIV data before committing to OSM... I can only conclude there is much work to be done, AGIV is far from recent concerning new built houses, and OSM itself has lots of issues regarding accuracy. I'm not too sure on the scientific significance of such a statistic either. I would more than love to see stats that compair quality of the entered addresses. (completeness , including postal code and other addr:* tags, number of corrections etc. ) I've been correcting a lot of mistakes and I start seeing a negative tendancy in it: The ones that have done a LOT of input but didn't care to quality check (validating even!) what they entered. I'll state this: I'm cleaning up far too much crap others leave behind, I'm far from perfect in my housenumbering too in this small village it's still a huge undertaking, especially on complex corners where some numbers of the same building belong to a different street. I have houses I've changed 3 times in a row after visiting. Moreover, I see some people tagging the next city name on a street, probably because they think it's in the next city, but OSM shouldn't be a guessing game or a race to enter the most addresses if that info is wrong. If someone has such an overpass query, I'm more than interested to see those results, the rest looks like bragging rights. The AGIV site is valuable , even though it's slow. It's great tool to verify what city a certain street belongs to. for example : http://www.openstreetmap.org/?lat=50.998941lon=4.426396zoom=18layers=M De Kleine Parijsstraat belongs to Zemst, not Mechelen. If you look this up in nominatim, it will tell you it's in Mechelen, which is totally wrong. You will not find this street using AGIV in Mechelen. But someone decided this was Hombeek(Mechelen) instead. So the borders of Zemst where wrong as well as this was used to determine these.The street above that Boterstraat can be found in Mechelen, not in Zemst. Thanks to AGIV, I'm more certain when those cases present. ( You can still see the old cached tile in some zoom levels) But then again, I saw AGIV containing WRONG housenumbers too. Verification in the field (twice) confirmed that what I had in mind was matching reality. So it's like a triple check: a) know the place b) visit it c) check with AGIV d) map a decent hous e) be complete, using the plugins to add Country/postal code/streetname to an address node so the data is easily searchable later. I really, really have to plea to everyone to enter _better_ data than more , just instead of looking at the sheer number of address info/nodes entered. It's quickly getting tired when I have to keep cleaning up behind the top providers. I'll get off the soapbox now. Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
There are good reasons for addr:street on individual building. A lot of programs (like some apps) don t show the associated street, or are difficult to find (Potlach). Not every contributor uses JOSM 2013/4/15 Marc Gemis marc.ge...@gmail.com Your complain about street being placed in wrong cities, is exactly why we should use associatedStreet relations instead of repeating addr:street addr:city over and over on individual buildings. In that case you only have to correct it once, on the relation, and the data is corrected. But I'll admit that I've been to lazy to add associatedStreets recently, as nobody seems to care. And the tools support outside JOSM is not that great. Repeating the street name twice is completely useless IMHO. I think it is possible to combine lot's of house numbers and accuracy, at least I hope that's what I leave behind :-) m On Mon, Apr 15, 2013 at 5:48 PM, Glenn Plas gl...@byte-consult.be wrote: On 04/15/2013 02:38 PM, JorenDC wrote: Hi, In December there was a thread (start: http://lists.openstreetmap.org/pipermail/talk-be/2012-December/003367.html) containing some numbers/stats. @Sander, *: is it possible to share your used method to pull these stats (or just pull them again) and publish them on a 'frequent' base (I'm not saying weekly, but what about +/- every 4 moths or so)? In that way we can see a bit of our progress regarding this 'project'. The method (overpass query) is mentioned in that link thread (Jo points it out). On the subject. I've been mapping a lot of housenumbers lately, verifying my data against AGIV data before committing to OSM... I can only conclude there is much work to be done, AGIV is far from recent concerning new built houses, and OSM itself has lots of issues regarding accuracy. I'm not too sure on the scientific significance of such a statistic either. I would more than love to see stats that compair quality of the entered addresses. (completeness , including postal code and other addr:* tags, number of corrections etc. ) I've been correcting a lot of mistakes and I start seeing a negative tendancy in it: The ones that have done a LOT of input but didn't care to quality check (validating even!) what they entered. I'll state this: I'm cleaning up far too much crap others leave behind, I'm far from perfect in my housenumbering too in this small village it's still a huge undertaking, especially on complex corners where some numbers of the same building belong to a different street. I have houses I've changed 3 times in a row after visiting. Moreover, I see some people tagging the next city name on a street, probably because they think it's in the next city, but OSM shouldn't be a guessing game or a race to enter the most addresses if that info is wrong. If someone has such an overpass query, I'm more than interested to see those results, the rest looks like bragging rights. The AGIV site is valuable , even though it's slow. It's great tool to verify what city a certain street belongs to. for example : http://www.openstreetmap.org/?lat=50.998941lon=4.426396zoom=18layers=M De Kleine Parijsstraat belongs to Zemst, not Mechelen. If you look this up in nominatim, it will tell you it's in Mechelen, which is totally wrong. You will not find this street using AGIV in Mechelen. But someone decided this was Hombeek(Mechelen) instead. So the borders of Zemst where wrong as well as this was used to determine these.The street above that Boterstraat can be found in Mechelen, not in Zemst. Thanks to AGIV, I'm more certain when those cases present. ( You can still see the old cached tile in some zoom levels) But then again, I saw AGIV containing WRONG housenumbers too. Verification in the field (twice) confirmed that what I had in mind was matching reality. So it's like a triple check: a) know the place b) visit it c) check with AGIV d) map a decent hous e) be complete, using the plugins to add Country/postal code/streetname to an address node so the data is easily searchable later. I really, really have to plea to everyone to enter _better_ data than more , just instead of looking at the sheer number of address info/nodes entered. It's quickly getting tired when I have to keep cleaning up behind the top providers. I'll get off the soapbox now. Glenn ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be ___ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
Re: [OSM-talk-be] Address stats in Belgium
It doesn't matter where they are , you still have to put those tags on a relation, so they better be in the correct city to start with ... Wrong postal codes, wrong city I would rather _NOT_ have wrong ones than the 'close enough for me' type of data. My point was introducing wrong data I admit I'm not a fan of the associatedStreet relation. I just recently learned it makes it even easier for some to f#ck up my work by 'correcting' a streetRelation that wasn't broken. Now it's even more easy to destroy 'en mass' in a single mouse click. There are some amazing josm plugins (Address plugin, adress mapcss etc) to help you do this without pain. We are all repeating 'building=yes' on a building, it's not because we put millions of this on the map that this tag get's to be valued less now than before? It's because it's needed and gives useful info. Why would your thoughts be any different for the addr:street tag that carry much more useful info than a mere 'building=yes'. So no, I don't think we are 'repeating' data, we are detailing it as such ... I sometimes believe I'm the only one in Belgium using OSM data in a non-mapping fashion like geocoding. There are working JOSM plugins that make sure you'll never ever have to type the streetname, you just select the named road + addr node + building and press = shift-T See the Terracer plugin. http://wiki.openstreetmap.org/wiki/JOSM/Plugins/Terracer But also, the FixAddresses plugin. http://wiki.openstreetmap.org/wiki/JOSM/Plugins/FixAddresses I just wished it supported AssociatedStreet relations, same goes for some of the finer mapCSS I see. I combined them all, and this gives me powerful view on the address situation in the target area. http://josm.openstreetmap.de/josmfile?page=Styles/AddressValidatorstyle https://github.com/simon04/coloured-addresses.mapcss/raw/master/dist/coloured-addresses.mapcss http://josm.openstreetmap.de/josmfile?page=Styles/Nonamestyle try them, you'll love the colors per street, very nice to spot problems on corners. Glenn On 04/15/2013 07:33 PM, Marc Gemis wrote: Your complain about street being placed in wrong cities, is exactly why we should use associatedStreet relations instead of repeating addr:street addr:city over and over on individual buildings. In that case you only have to correct it once, on the relation, and the data is corrected. But I'll admit that I've been to lazy to add associatedStreets recently, as nobody seems to care. And the tools support outside JOSM is not that great. Repeating the street name twice is completely useless IMHO. I think it is possible to combine lot's of house numbers and accuracy, at least I hope that's what I leave behind :-) m On Mon, Apr 15, 2013 at 5:48 PM, Glenn Plas gl...@byte-consult.be mailto:gl...@byte-consult.be wrote: On 04/15/2013 02:38 PM, JorenDC wrote: Hi, In December there was a thread (start: http://lists.openstreetmap.org/pipermail/talk-be/2012-December/003367.html) containing some numbers/stats. @Sander, *: is it possible to share your used method to pull these stats (or just pull them again) and publish them on a 'frequent' base (I'm not saying weekly, but what about +/- every 4 moths or so)? In that way we can see a bit of our progress regarding this 'project'. The method (overpass query) is mentioned in that link thread (Jo points it out). On the subject. I've been mapping a lot of housenumbers lately, verifying my data against AGIV data before committing to OSM... I can only conclude there is much work to be done, AGIV is far from recent concerning new built houses, and OSM itself has lots of issues regarding accuracy. I'm not too sure on the scientific significance of such a statistic either. I would more than love to see stats that compair quality of the entered addresses. (completeness , including postal code and other addr:* tags, number of corrections etc. ) I've been correcting a lot of mistakes and I start seeing a negative tendancy in it: The ones that have done a LOT of input but didn't care to quality check (validating even!) what they entered. I'll state this: I'm cleaning up far too much crap others leave behind, I'm far from perfect in my housenumbering too in this small village it's still a huge undertaking, especially on complex corners where some numbers of the same building belong to a different street. I have houses I've changed 3 times in a row after visiting. Moreover, I see some people tagging the next city name on a street, probably because they think it's in the next city, but OSM shouldn't be a guessing game or a race to enter the most addresses if that info is wrong. If someone has such an overpass query, I'm more than interested to see those results, the rest looks like bragging rights. The AGIV
Re: [OSM-talk-be] Address stats in Belgium
On Tue, Apr 16, 2013 at 6:46 AM, Jo winfi...@gmail.com wrote: I thought the consensus was that we repeat addr:street for each house but not the other information like addr:city, addr:postcode, addr:country and whatnot. Imagine what happens to the size of the DB (and all its derivatives like the planet files), if everybody starts doing that for each and every house/address in the world! Potlatch has its limitations, but by choosing to work with it, people indicate a willingness to live with those limitations. I can only hope the IDeditor will overcome those limitations one day and that all Potlatch users will migrate towards it when it does. Jo +1 We should not map for the renderer, nor for the (simpliest) editor. +1 also on not repeating the same data over and over. 2013/4/16 Marc Gemis marc.ge...@gmail.com I use most of these plugins. But recently I started using them less, because now I convert my GPX waypoints to OSM data points automatically. The only thing I do not do is repeating the postal code over and over, but since you insist, I'll do that from now on. :-) Can you look at e.g. http://www.openstreetmap.org/?lat=51.16279363632202lon=4.425258636474609zoom=16an area I mapped this winter. Please let me know if you think it can be improved. Yes, the city and the postal code are only in the relation. Yes, I know I should use building=house more consistently. Yes, I know I could add sidewalk, lit, parking lane tags as I did (already partially) in http://www.openstreetmap.org/?lat=51.134501695632935lon=4.385626316070557zoom=16 Let me know if you think the data can be improved, as I'm willing to improve my tagging habits. regards m On Mon, Apr 15, 2013 at 10:55 PM, Glenn Plas gl...@byte-consult.bewrote: It doesn't matter where they are , you still have to put those tags on a relation, so they better be in the correct city to start with ... Wrong postal codes, wrong city I would rather _NOT_ have wrong ones than the 'close enough for me' type of data. My point was introducing wrong data I admit I'm not a fan of the associatedStreet relation. I just recently learned it makes it even easier for some to f#ck up my work by 'correcting' a streetRelation that wasn't broken. Now it's even more easy to destroy 'en mass' in a single mouse click. There are some amazing josm plugins (Address plugin, adress mapcss etc) to help you do this without pain. We are all repeating 'building=yes' on a building, it's not because we put millions of this on the map that this tag get's to be valued less now than before? It's because it's needed and gives useful info. Why would your thoughts be any different for the addr:street tag that carry much more useful info than a mere 'building=yes'. So no, I don't think we are 'repeating' data, we are detailing it as such ... I sometimes believe I'm the only one in Belgium using OSM data in a non-mapping fashion like geocoding. There are working JOSM plugins that make sure you'll never ever have to type the streetname, you just select the named road + addr node + building and press = shift-T See the Terracer plugin. http://wiki.openstreetmap.org/wiki/JOSM/Plugins/Terracer But also, the FixAddresses plugin. http://wiki.openstreetmap.org/wiki/JOSM/Plugins/FixAddresses I just wished it supported AssociatedStreet relations, same goes for some of the finer mapCSS I see. I combined them all, and this gives me powerful view on the address situation in the target area. http://josm.openstreetmap.de/josmfile?page=Styles/AddressValidatorstyle https://github.com/simon04/coloured-addresses.mapcss/raw/master/dist/coloured-addresses.mapcss http://josm.openstreetmap.de/josmfile?page=Styles/Nonamestyle try them, you'll love the colors per street, very nice to spot problems on corners. Glenn On 04/15/2013 07:33 PM, Marc Gemis wrote: Your complain about street being placed in wrong cities, is exactly why we should use associatedStreet relations instead of repeating addr:street addr:city over and over on individual buildings. In that case you only have to correct it once, on the relation, and the data is corrected. But I'll admit that I've been to lazy to add associatedStreets recently, as nobody seems to care. And the tools support outside JOSM is not that great. Repeating the street name twice is completely useless IMHO. I think it is possible to combine lot's of house numbers and accuracy, at least I hope that's what I leave behind :-) m On Mon, Apr 15, 2013 at 5:48 PM, Glenn Plas gl...@byte-consult.bewrote: On 04/15/2013 02:38 PM, JorenDC wrote: Hi, In December there was a thread (start: http://lists.openstreetmap.org/pipermail/talk-be/2012-December/003367.html) containing some numbers/stats. @Sander, *: is it possible to share your used method to pull these stats (or just pull them again) and publish them on a 'frequent' base (I'm not saying
