Re: [Talk-us] identifying TIGER deserts
I know that the TIGER 2012 overlay available in JOSM has been very helpful and I think that a combination of a) increasing awareness of this overlay and b) making the overlay available in Potlatch would go a long way toward leveraging TIGER 2012. In general, I think it is not the geometry which is most impactful, but the increase in names and the decrease in spurious or spuriously connected ways that has the most impact. --ceyockey -Original Message- From: Ian Dees Sent: Jan 7, 2013 8:10 PM To: Alex Barth Cc: Ian Villeda , Richard Welty , "talk-us@openstreetmap.org Openstreetmap" Subject: Re: [Talk-us] identifying TIGER deserts On Mon, Jan 7, 2013 at 4:39 PM, Alex Barth wrote: On Jan 6, 2013, at 11:28 AM, Ian Dees wrote: > > I've done this before and it works great for ~z10+. Some more processing > needs to be done for the lower zoom levels so that we can display an overview > to make it easier to find problematic areas. > Any particular thoughts on what kind of processing? What about rasterizing the most prominent "color" (i.e. missing 2012 or 2007 changes) in a 1km² area and boiling it up like Mike's Green Means Go map? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 6, 2013, at 4:30 AM, Alex Barth wrote: > On Jan 4, 2013, at 3:33 PM, Richard Welty wrote: > >> On 1/4/13 1:31 PM, Michal Migurski wrote: >>> Interesting—how would you characterize "bad" roads? One characteristic of >>> crappy TIGER data is road wiggliness, is that what you mean? >> >> the tiger 2010/11/12 data is much better for many of the "bad tiger" areas. >> it'd be a bit of work to do >> a comparison of the data, but it might be able to yield the desired >> information. > > This is exactly the direction I'm thinking. What would it take to create a > TIGER 07 vs 12 comparison map and use this as a basis for identifying the > areas that need most attention? I've been talking to Ian over here and we > were thinking of intersecting OSM highway data with TIGER 12. This could be > achieved e. g. by overlaying a light, opaque OSM highway layer with a > contrasting TIGER layer, only exposing TIGER 12 geometry where it differs > from OSM. > > Thoughts? This approach makes sense. You're thinking to come up with a metric for mismatched area between the two datasets? Would it be enough to total it up into 1x1 km buckets or would you want it to be substantially more detailed? -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Mon, Jan 7, 2013 at 4:39 PM, Alex Barth wrote: > > On Jan 6, 2013, at 11:28 AM, Ian Dees wrote: > > > > I've done this before and it works great for ~z10+. Some more processing > needs to be done for the lower zoom levels so that we can display an > overview to make it easier to find problematic areas. > > > > Any particular thoughts on what kind of processing? What about rasterizing the most prominent "color" (i.e. missing 2012 or 2007 changes) in a 1km² area and boiling it up like Mike's Green Means Go map? ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 6, 2013, at 12:06 PM, Mike N wrote: > On 1/6/2013 11:37 AM, Mike N wrote: >> I like the idea of a TIGER 07 vs 12 comparison only being used as a >> trigger. If you compare TIGER 12 vs OSM, it will highlight all the >> TIGER 07 artifact roads that were removed because they were in error or >> no longer exist, but are still often in TIGER 12 even after a geometry >> refresh. > > On second thought, you'd need a 3-way compare to further exclude new TIGER 12 > features that have already been added to OSM. More data to process but an overlay/intersect approach could still work for ZL 10+. > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 6, 2013, at 11:28 AM, Ian Dees wrote: > > I've done this before and it works great for ~z10+. Some more processing > needs to be done for the lower zoom levels so that we can display an overview > to make it easier to find problematic areas. > Any particular thoughts on what kind of processing? Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On 1/6/2013 7:30 AM, Alex Barth wrote: This could be achieved e. g. by overlaying a light, opaque OSM highway layer with a contrasting TIGER layer, only exposing TIGER 12 geometry where it differs from OSM. Oops, forgot the link: http://greenvilleopenmap.info/TIGER12vsOSM.jpg Here is a TIGER 12 vs OSM comparison. They cyan inner arc exists in TIGER 12 but not OSM. However, it originally started out in TIGER 07 but was removed from OSM because it doesn't exist. One way around this is to do a TIGER 12 VS 07 compare to start with, then compare that to OSM to detect improved geometry or new roads. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On 1/6/2013 7:30 AM, Alex Barth wrote: This could be achieved e. g. by overlaying a light, opaque OSM highway layer with a contrasting TIGER layer, only exposing TIGER 12 geometry where it differs from OSM. Here is a TIGER 12 vs OSM comparison. They cyan inner arc exists in TIGER 12 but not OSM. However, it originally started out in TIGER 07 but was removed from OSM because it doesn't exist. One way around this is to do a TIGER 12 VS 07 compare to start with, then compare that to OSM to detect improved geometry or new roads. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On 1/6/2013 11:37 AM, Mike N wrote: I like the idea of a TIGER 07 vs 12 comparison only being used as a trigger. If you compare TIGER 12 vs OSM, it will highlight all the TIGER 07 artifact roads that were removed because they were in error or no longer exist, but are still often in TIGER 12 even after a geometry refresh. On second thought, you'd need a 3-way compare to further exclude new TIGER 12 features that have already been added to OSM. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On 1/6/2013 7:30 AM, Alex Barth wrote: This could be achieved e. g. by overlaying a light, opaque OSM highway layer with a contrasting TIGER layer, only exposing TIGER 12 geometry where it differs from OSM. I like the idea of a TIGER 07 vs 12 comparison only being used as a trigger. If you compare TIGER 12 vs OSM, it will highlight all the TIGER 07 artifact roads that were removed because they were in error or no longer exist, but are still often in TIGER 12 even after a geometry refresh. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Sun, Jan 6, 2013 at 6:30 AM, Alex Barth wrote: > > On Jan 4, 2013, at 3:33 PM, Richard Welty wrote: > > > On 1/4/13 1:31 PM, Michal Migurski wrote: > >> Interesting—how would you characterize "bad" roads? One characteristic > of crappy TIGER data is road wiggliness, is that what you mean? > > > > the tiger 2010/11/12 data is much better for many of the "bad tiger" > areas. it'd be a bit of work to do > > a comparison of the data, but it might be able to yield the desired > information. > > This is exactly the direction I'm thinking. What would it take to create a > TIGER 07 vs 12 comparison map and use this as a basis for identifying the > areas that need most attention? I've been talking to Ian over here and we > were thinking of intersecting OSM highway data with TIGER 12. This could be > achieved e. g. by overlaying a light, opaque OSM highway layer with a > contrasting TIGER layer, only exposing TIGER 12 geometry where it differs > from OSM. > > Thoughts? I've done this before and it works great for ~z10+. Some more processing needs to be done for the lower zoom levels so that we can display an overview to make it easier to find problematic areas. There tend to be small blobs of new geometry (new subdivisions or more accurate geometry) when comparing 07 and 12. It would be interesting to cluster these together and use them as an input to MapRoulette or a Task Manager to pick off the stuff that is brand new or completely changed between the two datasets. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 4, 2013, at 3:33 PM, Richard Welty wrote: > On 1/4/13 1:31 PM, Michal Migurski wrote: >> Interesting—how would you characterize "bad" roads? One characteristic of >> crappy TIGER data is road wiggliness, is that what you mean? > > the tiger 2010/11/12 data is much better for many of the "bad tiger" areas. > it'd be a bit of work to do > a comparison of the data, but it might be able to yield the desired > information. This is exactly the direction I'm thinking. What would it take to create a TIGER 07 vs 12 comparison map and use this as a basis for identifying the areas that need most attention? I've been talking to Ian over here and we were thinking of intersecting OSM highway data with TIGER 12. This could be achieved e. g. by overlaying a light, opaque OSM highway layer with a contrasting TIGER layer, only exposing TIGER 12 geometry where it differs from OSM. Thoughts? > > richard > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On 1/4/13 1:31 PM, Michal Migurski wrote: Interesting—how would you characterize "bad" roads? One characteristic of crappy TIGER data is road wiggliness, is that what you mean? the tiger 2010/11/12 data is much better for many of the "bad tiger" areas. it'd be a bit of work to do a comparison of the data, but it might be able to yield the desired information. richard ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 4, 2013, at 1:33 AM, Alex Barth wrote: > On Jan 3, 2013, at 7:06 PM, Michal Migurski wrote: > >> Sounds great, I'm actually in the middle of publishing a followup to the >> green-means-go map after ploughing through the full planet history dump. > > Great Tue 1/8, 5PM Eastern. My handle is lx_barth, I'll be on #osm-dev at the > same time (alexb). Anybody who's interested here in identifying TIGER deserts > is welcome to join. > > I just looked at your new green-means-go map. There is the gap that I'm > seeing between what you're doing and what Ruben, Ian and I'd like to do: > we're particularly interested in places where the TIGER geography is off. I. > e. where roads are not where they should be. > > After a first test here with Virginia [1] we saw that TIGER deserts do not > necessarily equal bad road geography. So we're right now particularly > interested in methods for identifying bad TIGER road geography. > > I'm thinking such a "Bad TIGER Roads Map" could be a valuable community > resource to see where stuff is worst and needs fixing that we could maintain > over time. Interesting—how would you characterize "bad" roads? One characteristic of crappy TIGER data is road wiggliness, is that what you mean? -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 3, 2013, at 7:06 PM, Michal Migurski wrote: > On Jan 3, 2013, at 3:25 PM, Alex Barth wrote: >> >> Mike (and anyone else interested in TIGER deserts): >> >> Ruben, Ian and I would love to touch base over voice and see where we can >> cooperate. Skype or Google Hangout would be great. Could Tuesday 5PM Eastern >> work? > > > Sounds great, I'm actually in the middle of publishing a followup to the > green-means-go map after ploughing through the full planet history dump. Great Tue 1/8, 5PM Eastern. My handle is lx_barth, I'll be on #osm-dev at the same time (alexb). Anybody who's interested here in identifying TIGER deserts is welcome to join. I just looked at your new green-means-go map. There is the gap that I'm seeing between what you're doing and what Ruben, Ian and I'd like to do: we're particularly interested in places where the TIGER geography is off. I. e. where roads are not where they should be. After a first test here with Virginia [1] we saw that TIGER deserts do not necessarily equal bad road geography. So we're right now particularly interested in methods for identifying bad TIGER road geography. I'm thinking such a "Bad TIGER Roads Map" could be a valuable community resource to see where stuff is worst and needs fixing that we could maintain over time. [1] http://bl.ocks.org/d/ff2607349777ddee9181/ > > -mike. > > > michal migurski- contact info and pgp key: > sf/cahttp://mike.teczno.com/contact.html > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Jan 3, 2013, at 3:25 PM, Alex Barth wrote: > On Dec 20, 2012, at 2:19 PM, Michal Migurski wrote: > >> On Dec 19, 2012, at 5:39 PM, Ruben Lopez Mendoza wrote: >>> >>> A few obvious steps stand out, including rendering a national version of >>> these maps. I'd also love to figure out whether it makes sense to join >>> forces with Mike Migurski's Green Means Go map. >> >> Joining forces would be fun! > > Mike (and anyone else interested in TIGER deserts): > > Ruben, Ian and I would love to touch base over voice and see where we can > cooperate. Skype or Google Hangout would be great. Could Tuesday 5PM Eastern > work? Sounds great, I'm actually in the middle of publishing a followup to the green-means-go map after ploughing through the full planet history dump. Skype would be my strong preference. -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Dec 20, 2012, at 2:19 PM, Michal Migurski wrote: > On Dec 19, 2012, at 5:39 PM, Ruben Lopez Mendoza wrote: >> >> A few obvious steps stand out, including rendering a national version of >> these maps. I'd also love to figure out whether it makes sense to join >> forces with Mike Migurski's Green Means Go map. > > Joining forces would be fun! Mike (and anyone else interested in TIGER deserts): Ruben, Ian and I would love to touch base over voice and see where we can cooperate. Skype or Google Hangout would be great. Could Tuesday 5PM Eastern work? > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Russ Nelson writes: > Michal Migurski writes: > > Also, what's the deal with the Massachusetts TIGER import? > > Massachusetts had already made an improved version of the TIGER data, > so the decision was made to import that instead. I'm not sure it's a version of the TIGER data, but early on (I have no reason to doubt Russ's date), road data was imported from MassGIS. The origin of the data was "EOT", at the time Executive Office of Transportation, and now "MassDOT", more conventionally named. This data is generally of quite high quality. There were a few issues, now almost entirely behind us: Oneway roads were tagged oneway=yes, but the direction was unknown. This was apparently due to licensing confusion, but for the most part humans have fixed up the directions and removed the ~fixme:unreviewed-oneway tag. A few towns weren't imported. I think they either have been now, with manual merging, or mostly hand mapped. There are a very small number of errors, to the point when if you find an error in the massgis/eot data it's remarkable. (Of course there are new roads built since the dataset; that's not really an error.) In my town I have found exactly one issue: a pair of phantom road and a misnamed real road nearby. One can nitpick about the geometry quality at the junction of exit ramps and regular roads. Other than that, it's excellent. That said, the notion of identifying areas that haven't had the road geometry touched since the import. But, one shouldn't conclude that not touching the roads means they are wrong. I've generally only touched roads in my town to add connections to driveways. But, I've verified that many more are right, by watching a garmin with osm while driving, and looking at imagery in JOSM. On the other hand, a whole town with almost no changes probably has had little human attention. pgpwxWk8NFro2.pgp Description: PGP signature ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Michal Migurski writes: > Ah, good to know. Any idea what the approximate date and importing > account were? "MassGIS Import" somewhere around 10/13/07. -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Dec 22, 2012, at 7:01 PM, Russ Nelson wrote: > Michal Migurski writes: >> Also, what's the deal with the Massachusetts TIGER import? > > Massachusetts had already made an improved version of the TIGER data, > so the decision was made to import that instead. Ah, good to know. Any idea what the approximate date and importing account were? -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Michal Migurski writes: > Also, what's the deal with the Massachusetts TIGER import? Massachusetts had already made an improved version of the TIGER data, so the decision was made to import that instead. -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Dec 20, 2012, at 11:19 AM, Michal Migurski wrote: >> next steps >> >> A few obvious steps stand out, including rendering a national version of >> these maps. I'd also love to figure out whether it makes sense to join >> forces with Mike Migurski's Green Means Go map. >> >> It'd also be interesting to use a smarter Osmosis command (and the full >> history file) to limit the input data to just those ways created by the 3 >> primary user accounts associated with the TIGER >> import:https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc#L45 >> (h/t Migurski) > > Joining forces would be fun! > > For the past two days, I've been processing the Full History file into a form > that's easier to cope with for a big categorization run. I want to look at > the individual nodes, using their associated changesets to pick up on > non-import edits that fell through the cracks when I looked only at OSM ways. > > Here's where I'm at now: > http://www.openstreetmap.us/~migurski/TIGER-Raster/nodes/ I've updated the page above with a preliminary rendering showing non-TIGER node density, preview and full GeoTIFF here: http://www.openstreetmap.us/~migurski/TIGER-Raster/nodes/non-tiger-uids.jpg http://www.openstreetmap.us/~migurski/TIGER-Raster/nodes/out-full-history-uids-no.tif.bz2 I'm processing the full history planet file a second time now, paying attention to just the nodes that make up ways with highway tags, since I'm noticing a lot of large non-road imports in this dataset. Also, what's the deal with the Massachusetts TIGER import? When did it happen? The state is conspicuously missing when I use the Mapquest TIGER edits case statement: (case when osm_uid = '7168' -- DaveHansenTiger and osm_timestamp::timestamp >= '2007-08-03'::timestamp and osm_timestamp::timestamp <= '2008-05-04'::timestamp then 0 when osm_uid = '15169' -- Milenko and osm_timestamp::timestamp >= '2007-10-29'::timestamp and osm_timestamp::timestamp <= '2007-12-12'::timestamp then 0 when osm_uid = '20587' -- balrog-kun and osm_timestamp::timestamp >= '2010-03-21'::timestamp and osm_timestamp::timestamp <= '2010-04-08'::timestamp and osm_version::int < 3 -- maybe someone else edited between import and name expansion then 0 else 1 end) as is_touched (Github currently dead but it's from https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc) -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Dec 19, 2012, at 5:39 PM, Ruben Lopez Mendoza wrote: > We've worked out some kinks of our first attempt to identify TIGER deserts, > and produced maps a couple maps along the way. > > … > I've captured all of this work - the postgres functions, shapefiles, and > tilemill projects in the a repo here:https://github.com/Rub21/tiger-deserts +1. > next steps > > A few obvious steps stand out, including rendering a national version of > these maps. I'd also love to figure out whether it makes sense to join forces > with Mike Migurski's Green Means Go map. > > It'd also be interesting to use a smarter Osmosis command (and the full > history file) to limit the input data to just those ways created by the 3 > primary user accounts associated with the TIGER > import:https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc#L45 > (h/t Migurski) Joining forces would be fun! For the past two days, I've been processing the Full History file into a form that's easier to cope with for a big categorization run. I want to look at the individual nodes, using their associated changesets to pick up on non-import edits that fell through the cracks when I looked only at OSM ways. Here's where I'm at now: http://www.openstreetmap.us/~migurski/TIGER-Raster/nodes/ -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Hello all, Ruben led out MapBox's analysis, the results of which you can see in his previous message. While these maps do a fairly good job of identifying areas that have and have not been edited, we're not entirely satisfied with their ability to identify where TIGER ways just don't represent the actual geographic location of roads on the ground. Another thought we've had is to do some flavor of comparative analysis between 2007 TIGER and 2012 TIGER. This might me clipping 2007 to the 2012 dataset to identify where roads do not overlap (and then perhaps some raster/binning to identify problem areas). Just wondering what other ideas might be out there - hit me up with your ideas. thanks! -- ian villeda mapbox | developmentseed https://twitter.com/ian_villeda On Wednesday, December 19, 2012 at 8:39 PM, Ruben Lopez Mendoza wrote: > > We've worked out some kinks of our first attempt to identify TIGER deserts, > and produced maps a couple maps along the way. > > > Average version of all highway =* ways > https://tiles.mapbox.com/ruben/map/map-badetj1b#7.00/37.927/-78.466 > > > Percent version 1, highwahy=* ways per gridcell > http://tiles.mapbox.com/ruben/map/map-4t3notfw#9.00/37.2805/-78.2528 > > > Comparison: http://bl.ocks.org/d/ff2607349777ddee9181/ > > > As input we used a database of all highway=* ways and their version number > created using Osmosis and a grid generated from a Natural Earth shapefile. > > > I created a postgres function that populates a new table called tiger_grid. > For each cell, we calculated each of the following: > > amount_version1 > amount version2 > percentage _version1 > percentage version2 > average_version > > > I've captured all of this work - the postgres functions, shapefiles, and > tilemill projects in the a repo here:https://github.com/Rub21/tiger-deserts > > next steps > > A few obvious steps stand out, including rendering a national version of > these maps. I'd also love to figure out whether it makes sense to join forces > with Mike Migurski's Green Means Go map. > > > It'd also be interesting to use a smarter Osmosis command (and the full > history file) to limit the input data to just those ways created by the 3 > primary user accounts associated with the TIGER > import:https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc#L45 > (h/t Migurski) > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org (mailto:Talk-us@openstreetmap.org) > http://lists.openstreetmap.org/listinfo/talk-us > > ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Some context for the query below: https://gist.github.com/4291608 I've been using this projection for everything, a spherical Albers to make it possible to generate a slippy map in a conic projection: +proj=aea +lat_1=29.5 +lat_2=45.5 +lat_0=23 +lon_0=-96 +x_0=0 +y_0=0 +a=6378137 +b=6378137 +nadgrids=@null +units=m +no_defs -mike. On Dec 14, 2012, at 9:47 PM, Michal Migurski wrote: > Are you guys using the way version number to determine TIGERness? This from > Andy Allan might be more effective: > >-- >-- TIGER edits case statement from >-- > https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc >-- >(case when osm_uid = '7168' -- DaveHansenTiger >and osm_timestamp::timestamp >= '2007-08-03'::timestamp >and osm_timestamp::timestamp <= '2008-05-04'::timestamp > then 0 > when osm_uid = '15169' -- Milenko >and osm_timestamp::timestamp >= '2007-10-29'::timestamp >and osm_timestamp::timestamp <= '2007-12-12'::timestamp > then 0 > when osm_uid = '20587' -- balrog-kun >and osm_timestamp::timestamp >= '2010-03-21'::timestamp >and osm_timestamp::timestamp <= '2010-04-08'::timestamp >and osm_version::int < 3 -- maybe someone else edited between > import and name expansion > then 0 > else 1 end) as is_touched > > Number of distinct users who've edited in a given area would also be a useful > metric, though incomplete without reference to the full history file. > > -mike. > > On Dec 14, 2012, at 7:45 PM, Alex Barth wrote: > >> >> Before I log off for tonight, let me share this screenshot I just got from >> Ruben (cc'ed) showing his progress on creating a slippy TIGER deserts map, >> pretty much using Martijn's approach of binned average geometry versions. >> These are first steps and there are still some math problems to iron out, >> expect an update next week. >> >> http://cl.ly/image/0x3h0s1H1v0k >> >> On Dec 7, 2012, at 1:00 PM, Ian Villeda wrote: >> >>> >>> Hey team, >>> >>> Over here at MapBox we were inspired by Martjin's great post on identifying >>> TIGER deserts[1], so we're attempting to effectively identify TIGER deserts >>> for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). >>> The goal is to create a slippy map from zoom level 2 to 15(?) such that >>> it's easy to find the places in need of the most love. >>> >>> We have a osmium-genereated postigs database with all ways tagged as >>> highway=* and their version number. The plan to run the analysis against a >>> 5x5 km grid, although we will probably play with the size of the gridcell. >>> Another option might be to conduct the analysis and varying levels of >>> granularity. >>> >>> We're still consiering different methods of analysis - the goal being to >>> normalize way density to display better identify TIGER deserts. We'd would >>> love your thoughts / input / reactions. A few ideas we've tossed around / >>> are activley pursuing: >>> >>> - percent of version 1 and/or 2 ways in each gridcell >>> - average number of version 1/2 ways per gridcell >>> - using a threshold value(s) similiar to Martjin's approach >>> - [your ideas here] >>> >>> Ultimately we'd like showcase the rendered results in site that would make >>> editing TIGER deserts easy. For now though this is just a heads up and an >>> invitation if you have any thoughts on how better to identify TIGER >>> deserts. >>> >>> thanks! >>> >>> [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ >>> -- >>> ian villeda >>> mapbox | developmentseed >>> https://twitter.com/ian_villeda >>> >>> >>> ___ >>> Talk-us mailing list >>> Talk-us@openstreetmap.org >>> http://lists.openstreetmap.org/listinfo/talk-us >> >> Alex Barth >> http://twitter.com/lxbarth >> tel (+1) 202 250 3633 >> >> >> >> >> >> ___ >> Talk-us mailing list >> Talk-us@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-us >> > > > michal migurski- contact info and pgp key: > sf/cahttp://mike.teczno.com/contact.html > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us > michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Are you guys using the way version number to determine TIGERness? This from Andy Allan might be more effective: -- -- TIGER edits case statement from -- https://github.com/MapQuest/TIGER-Edited-map/blob/master/inc/layer-tiger.xml.inc -- (case when osm_uid = '7168' -- DaveHansenTiger and osm_timestamp::timestamp >= '2007-08-03'::timestamp and osm_timestamp::timestamp <= '2008-05-04'::timestamp then 0 when osm_uid = '15169' -- Milenko and osm_timestamp::timestamp >= '2007-10-29'::timestamp and osm_timestamp::timestamp <= '2007-12-12'::timestamp then 0 when osm_uid = '20587' -- balrog-kun and osm_timestamp::timestamp >= '2010-03-21'::timestamp and osm_timestamp::timestamp <= '2010-04-08'::timestamp and osm_version::int < 3 -- maybe someone else edited between import and name expansion then 0 else 1 end) as is_touched Number of distinct users who've edited in a given area would also be a useful metric, though incomplete without reference to the full history file. -mike. On Dec 14, 2012, at 7:45 PM, Alex Barth wrote: > > Before I log off for tonight, let me share this screenshot I just got from > Ruben (cc'ed) showing his progress on creating a slippy TIGER deserts map, > pretty much using Martijn's approach of binned average geometry versions. > These are first steps and there are still some math problems to iron out, > expect an update next week. > > http://cl.ly/image/0x3h0s1H1v0k > > On Dec 7, 2012, at 1:00 PM, Ian Villeda wrote: > >> >> Hey team, >> >> Over here at MapBox we were inspired by Martjin's great post on identifying >> TIGER deserts[1], so we're attempting to effectively identify TIGER deserts >> for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). >> The goal is to create a slippy map from zoom level 2 to 15(?) such that it's >> easy to find the places in need of the most love. >> >> We have a osmium-genereated postigs database with all ways tagged as >> highway=* and their version number. The plan to run the analysis against a >> 5x5 km grid, although we will probably play with the size of the gridcell. >> Another option might be to conduct the analysis and varying levels of >> granularity. >> >> We're still consiering different methods of analysis - the goal being to >> normalize way density to display better identify TIGER deserts. We'd would >> love your thoughts / input / reactions. A few ideas we've tossed around / >> are activley pursuing: >> >> - percent of version 1 and/or 2 ways in each gridcell >> - average number of version 1/2 ways per gridcell >> - using a threshold value(s) similiar to Martjin's approach >> - [your ideas here] >> >> Ultimately we'd like showcase the rendered results in site that would make >> editing TIGER deserts easy. For now though this is just a heads up and an >> invitation if you have any thoughts on how better to identify TIGER deserts. >> >> thanks! >> >> [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ >> -- >> ian villeda >> mapbox | developmentseed >> https://twitter.com/ian_villeda >> >> >> ___ >> Talk-us mailing list >> Talk-us@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-us > > Alex Barth > http://twitter.com/lxbarth > tel (+1) 202 250 3633 > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us > michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Before I log off for tonight, let me share this screenshot I just got from Ruben (cc'ed) showing his progress on creating a slippy TIGER deserts map, pretty much using Martijn's approach of binned average geometry versions. These are first steps and there are still some math problems to iron out, expect an update next week. http://cl.ly/image/0x3h0s1H1v0k On Dec 7, 2012, at 1:00 PM, Ian Villeda wrote: > > Hey team, > > Over here at MapBox we were inspired by Martjin's great post on identifying > TIGER deserts[1], so we're attempting to effectively identify TIGER deserts > for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). > The goal is to create a slippy map from zoom level 2 to 15(?) such that it's > easy to find the places in need of the most love. > > We have a osmium-genereated postigs database with all ways tagged as > highway=* and their version number. The plan to run the analysis against a > 5x5 km grid, although we will probably play with the size of the gridcell. > Another option might be to conduct the analysis and varying levels of > granularity. > > We're still consiering different methods of analysis - the goal being to > normalize way density to display better identify TIGER deserts. We'd would > love your thoughts / input / reactions. A few ideas we've tossed around / are > activley pursuing: > > - percent of version 1 and/or 2 ways in each gridcell > - average number of version 1/2 ways per gridcell > - using a threshold value(s) similiar to Martjin's approach > - [your ideas here] > > Ultimately we'd like showcase the rendered results in site that would make > editing TIGER deserts easy. For now though this is just a heads up and an > invitation if you have any thoughts on how better to identify TIGER deserts. > > thanks! > > [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ > -- > ian villeda > mapbox | developmentseed > https://twitter.com/ian_villeda > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Dec 7, 2012, at 1:25 PM, Paul Norman wrote: >> From: Michal Migurski [mailto:m...@teczno.com] >> Subject: Re: [Talk-us] identifying TIGER deserts >> >> That's great! >> >> I've been thinking about some of the same things, inspired by Dennis >> Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 >> kilometer squares, measuring OSM user count (post-import) vs. >> possibility for improvement between 2007 and 2012 TIGER/Line data: >> >> http://mike.teczno.com/img/osm-users-imports-2012-09.png >> - Blue means 2012 data would be an improvement > > How do you identify this? I know that some of the biggest improvements I've > made to TIGER data in remote areas was to delete half of the data. If the > TIGER2012 data hasn't changed a naive comparison will say that the 2012 data > has twice the road length and is better, yet the OSM data is really better > by virtue of paper roads having been deleted. I identify it purely based on length, so yeah. If someone did nothing but delete TIGER roads then it would look like 2012 would be a huge improvement in an available area. If someone touched any one of the TIGER roads then it should cause that area to appear as a red/edited. -mike. michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Something else that would be cool for the US OSM server is a tilecache proxy of publicly available, kosher data sources On Fri, Dec 7, 2012 at 5:07 PM, Michal Migurski wrote: > Thanks! > > I agree about a regularly-updated map. For this one, I used the up-to-date > osm2pgsql database that Ian Dees is maintaining on the OSM-US server we > have living in Oregon, an awesome resource for USian OSM things. =) > > I re-projected everything to spherical albers, and then iterated over > every grid square in the raster and did a simple Postgres SUM(ST_Length) > query for the roads in that area. The data was written out as a set of > Float32 rasters, which I later combined with Numpy to make pretty pictures. > > -mike. > > On Dec 7, 2012, at 1:09 PM, Alex Barth wrote: > > > Mike - > > > > AWESOME looking map and similar to what Ian and Ruben are working on. > Would you be open to share how you built that map? How hard (time > consuming) is it to regenerate? > > > > I'd love to get a regularly updated map out there showing TIGER deserts. > We'd use that ourselves to start fixing some of the worst areas in terms of > geographic accuracy. But obviously such a map would be a good thing to have > for the wider community. > > > > On Dec 7, 2012, at 1:32 PM, Michal Migurski wrote: > > > >> That's great! > >> > >> I've been thinking about some of the same things, inspired by Dennis > Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 kilometer > squares, measuring OSM user count (post-import) vs. possibility for > improvement between 2007 and 2012 TIGER/Line data: > >> > >> http://mike.teczno.com/img/osm-users-imports-2012-09.png > >> - Blue means 2012 data would be an improvement > >> - Red/Yellow means there are active OSM users there > >> > >> I used an Albers equal-area projection to ensure correct length and > density regardless of latitude, and my intent for this is to produce > county-by-county views that GIS managers in different jurisdictions can use > to understand whether their data is worth important, and what OSM users > they'd need to contact in order to be successful. > >> > >> I have a slippy map version of this but the projection registration is > off by a few hundred meters; need to figure out why that is and fix it. > >> > >> -mike. > >> > >> On Dec 7, 2012, at 10:00 AM, Ian Villeda wrote: > >> > >>> > >>> Hey team, > >>> > >>> Over here at MapBox we were inspired by Martjin's great post on > identifying TIGER deserts[1], so we're attempting to effectively identify > TIGER deserts for the rest of the continental U.S. (sorry Hawaii + Alaska, > you're next). The goal is to create a slippy map from zoom level 2 to 15(?) > such that it's easy to find the places in need of the most love. > >>> > >>> We have a osmium-genereated postigs database with all ways tagged as > highway=* and their version number. The plan to run the analysis against a > 5x5 km grid, although we will probably play with the size of the gridcell. > Another option might be to conduct the analysis and varying levels of > granularity. > >>> > >>> We're still consiering different methods of analysis - the goal being > to normalize way density to display better identify TIGER deserts. We'd > would love your thoughts / input / reactions. A few ideas we've tossed > around / are activley pursuing: > >>> > >>> - percent of version 1 and/or 2 ways in each gridcell > >>> - average number of version 1/2 ways per gridcell > >>> - using a threshold value(s) similiar to Martjin's approach > >>> - [your ideas here] > >>> > >>> Ultimately we'd like showcase the rendered results in site that would > make editing TIGER deserts easy. For now though this is just a heads up and > an invitation if you have any thoughts on how better to identify TIGER > deserts. > >>> > >>> thanks! > >>> > >>> [1]: > http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ > >>> -- > >>> ian villeda > >>> mapbox | developmentseed > >>> https://twitter.com/ian_villeda > >>> > >>> > >>> ___ > >>> Talk-us mailing list > >>> Talk-us@openstreetmap.org > >>> http://lists.openstreetmap.org/listinfo/talk-us > >> > >> > >> michal migurski- contact info and pgp key: > >> sf/cahttp://mike.teczno.com/contact.html > >> > >> > >> > >> > >> > >> ___ > >> Talk-us mailing list > >> Talk-us@openstreetmap.org > >> http://lists.openstreetmap.org/listinfo/talk-us > > > > Alex Barth > > http://twitter.com/lxbarth > > tel (+1) 202 250 3633 > > > > > > > > > > > > ___ > > Talk-us mailing list > > Talk-us@openstreetmap.org > > http://lists.openstreetmap.org/listinfo/talk-us > > > > > michal migurski- contact info and pgp key: > sf/cahttp://mike.teczno.com/contact.html > > > > >
Re: [Talk-us] identifying TIGER deserts
Thanks! I agree about a regularly-updated map. For this one, I used the up-to-date osm2pgsql database that Ian Dees is maintaining on the OSM-US server we have living in Oregon, an awesome resource for USian OSM things. =) I re-projected everything to spherical albers, and then iterated over every grid square in the raster and did a simple Postgres SUM(ST_Length) query for the roads in that area. The data was written out as a set of Float32 rasters, which I later combined with Numpy to make pretty pictures. -mike. On Dec 7, 2012, at 1:09 PM, Alex Barth wrote: > Mike - > > AWESOME looking map and similar to what Ian and Ruben are working on. Would > you be open to share how you built that map? How hard (time consuming) is it > to regenerate? > > I'd love to get a regularly updated map out there showing TIGER deserts. We'd > use that ourselves to start fixing some of the worst areas in terms of > geographic accuracy. But obviously such a map would be a good thing to have > for the wider community. > > On Dec 7, 2012, at 1:32 PM, Michal Migurski wrote: > >> That's great! >> >> I've been thinking about some of the same things, inspired by Dennis >> Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 kilometer >> squares, measuring OSM user count (post-import) vs. possibility for >> improvement between 2007 and 2012 TIGER/Line data: >> >> http://mike.teczno.com/img/osm-users-imports-2012-09.png >> - Blue means 2012 data would be an improvement >> - Red/Yellow means there are active OSM users there >> >> I used an Albers equal-area projection to ensure correct length and density >> regardless of latitude, and my intent for this is to produce >> county-by-county views that GIS managers in different jurisdictions can use >> to understand whether their data is worth important, and what OSM users >> they'd need to contact in order to be successful. >> >> I have a slippy map version of this but the projection registration is off >> by a few hundred meters; need to figure out why that is and fix it. >> >> -mike. >> >> On Dec 7, 2012, at 10:00 AM, Ian Villeda wrote: >> >>> >>> Hey team, >>> >>> Over here at MapBox we were inspired by Martjin's great post on identifying >>> TIGER deserts[1], so we're attempting to effectively identify TIGER deserts >>> for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). >>> The goal is to create a slippy map from zoom level 2 to 15(?) such that >>> it's easy to find the places in need of the most love. >>> >>> We have a osmium-genereated postigs database with all ways tagged as >>> highway=* and their version number. The plan to run the analysis against a >>> 5x5 km grid, although we will probably play with the size of the gridcell. >>> Another option might be to conduct the analysis and varying levels of >>> granularity. >>> >>> We're still consiering different methods of analysis - the goal being to >>> normalize way density to display better identify TIGER deserts. We'd would >>> love your thoughts / input / reactions. A few ideas we've tossed around / >>> are activley pursuing: >>> >>> - percent of version 1 and/or 2 ways in each gridcell >>> - average number of version 1/2 ways per gridcell >>> - using a threshold value(s) similiar to Martjin's approach >>> - [your ideas here] >>> >>> Ultimately we'd like showcase the rendered results in site that would make >>> editing TIGER deserts easy. For now though this is just a heads up and an >>> invitation if you have any thoughts on how better to identify TIGER >>> deserts. >>> >>> thanks! >>> >>> [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ >>> -- >>> ian villeda >>> mapbox | developmentseed >>> https://twitter.com/ian_villeda >>> >>> >>> ___ >>> Talk-us mailing list >>> Talk-us@openstreetmap.org >>> http://lists.openstreetmap.org/listinfo/talk-us >> >> >> michal migurski- contact info and pgp key: >> sf/cahttp://mike.teczno.com/contact.html >> >> >> >> >> >> ___ >> Talk-us mailing list >> Talk-us@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-us > > Alex Barth > http://twitter.com/lxbarth > tel (+1) 202 250 3633 > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us > michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
On Fri, Dec 7, 2012 at 1:25 PM, Paul Norman wrote: > > > How do you identify this? I know that some of the biggest improvements I've > made to TIGER data in remote areas was to delete half of the data. If the > TIGER2012 data hasn't changed a naive comparison will say that the 2012 > data > has twice the road length and is better, yet the OSM data is really better > by virtue of paper roads having been deleted. > > > full ack. Have done that myself in areas I visit and with the help of aerial pics. in many places this can reduce the total road length by > 50% and most has to be downgraded to track. > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us > ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
> From: Michal Migurski [mailto:m...@teczno.com] > Subject: Re: [Talk-us] identifying TIGER deserts > > That's great! > > I've been thinking about some of the same things, inspired by Dennis > Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 > kilometer squares, measuring OSM user count (post-import) vs. > possibility for improvement between 2007 and 2012 TIGER/Line data: > > http://mike.teczno.com/img/osm-users-imports-2012-09.png > - Blue means 2012 data would be an improvement How do you identify this? I know that some of the biggest improvements I've made to TIGER data in remote areas was to delete half of the data. If the TIGER2012 data hasn't changed a naive comparison will say that the 2012 data has twice the road length and is better, yet the OSM data is really better by virtue of paper roads having been deleted. ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Mike - AWESOME looking map and similar to what Ian and Ruben are working on. Would you be open to share how you built that map? How hard (time consuming) is it to regenerate? I'd love to get a regularly updated map out there showing TIGER deserts. We'd use that ourselves to start fixing some of the worst areas in terms of geographic accuracy. But obviously such a map would be a good thing to have for the wider community. On Dec 7, 2012, at 1:32 PM, Michal Migurski wrote: > That's great! > > I've been thinking about some of the same things, inspired by Dennis > Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 kilometer > squares, measuring OSM user count (post-import) vs. possibility for > improvement between 2007 and 2012 TIGER/Line data: > > http://mike.teczno.com/img/osm-users-imports-2012-09.png > - Blue means 2012 data would be an improvement > - Red/Yellow means there are active OSM users there > > I used an Albers equal-area projection to ensure correct length and density > regardless of latitude, and my intent for this is to produce county-by-county > views that GIS managers in different jurisdictions can use to understand > whether their data is worth important, and what OSM users they'd need to > contact in order to be successful. > > I have a slippy map version of this but the projection registration is off by > a few hundred meters; need to figure out why that is and fix it. > > -mike. > > On Dec 7, 2012, at 10:00 AM, Ian Villeda wrote: > >> >> Hey team, >> >> Over here at MapBox we were inspired by Martjin's great post on identifying >> TIGER deserts[1], so we're attempting to effectively identify TIGER deserts >> for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). >> The goal is to create a slippy map from zoom level 2 to 15(?) such that it's >> easy to find the places in need of the most love. >> >> We have a osmium-genereated postigs database with all ways tagged as >> highway=* and their version number. The plan to run the analysis against a >> 5x5 km grid, although we will probably play with the size of the gridcell. >> Another option might be to conduct the analysis and varying levels of >> granularity. >> >> We're still consiering different methods of analysis - the goal being to >> normalize way density to display better identify TIGER deserts. We'd would >> love your thoughts / input / reactions. A few ideas we've tossed around / >> are activley pursuing: >> >> - percent of version 1 and/or 2 ways in each gridcell >> - average number of version 1/2 ways per gridcell >> - using a threshold value(s) similiar to Martjin's approach >> - [your ideas here] >> >> Ultimately we'd like showcase the rendered results in site that would make >> editing TIGER deserts easy. For now though this is just a heads up and an >> invitation if you have any thoughts on how better to identify TIGER deserts. >> >> thanks! >> >> [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ >> -- >> ian villeda >> mapbox | developmentseed >> https://twitter.com/ian_villeda >> >> >> ___ >> Talk-us mailing list >> Talk-us@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-us > > > michal migurski- contact info and pgp key: > sf/cahttp://mike.teczno.com/contact.html > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us Alex Barth http://twitter.com/lxbarth tel (+1) 202 250 3633 ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
Absolutely, yeah. I chose a raster approach to the data I was working with because I imagined it would be easiest to adapt to arbitrary jurisdictions, whether adopt-a-pixel or masked by a county boundary. Easy to track buckets of data over time that way, too. -mike. On Dec 7, 2012, at 10:38 AM, Martijn van Exel wrote: > This would also fit in really well with the data steward notion. Part > of the idea that we've been toying with for this is to have data > dashboards for the stewarded areas to support targeted fixing / > improving and to build a stronger local discussion and community > around data > information > knowledge. > > On Fri, Dec 7, 2012 at 11:32 AM, Michal Migurski wrote: >> my intent for this is to produce county-by-county views that GIS managers in >> different jurisdictions can use to understand whether their data is worth > > > > -- > Martijn van Exel > http://oegeo.wordpress.com/ > http://openstreetmap.us/ > michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
I was just working on some user stories for the data steward idea, feel free to comment / add. I hope this will fit into the discussion around TIGER deserts and updates. https://docs.google.com/presentation/d/1mgKHtP9drx_rCXy_y3PPx_ZIBRLs1ASViorzpeOUZ0o/edit On Fri, Dec 7, 2012 at 11:38 AM, Martijn van Exel wrote: > This would also fit in really well with the data steward notion. Part > of the idea that we've been toying with for this is to have data > dashboards for the stewarded areas to support targeted fixing / > improving and to build a stronger local discussion and community > around data > information > knowledge. > > On Fri, Dec 7, 2012 at 11:32 AM, Michal Migurski wrote: >> my intent for this is to produce county-by-county views that GIS managers in >> different jurisdictions can use to understand whether their data is worth > > > > -- > Martijn van Exel > http://oegeo.wordpress.com/ > http://openstreetmap.us/ -- Martijn van Exel http://oegeo.wordpress.com/ http://openstreetmap.us/ ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
This would also fit in really well with the data steward notion. Part of the idea that we've been toying with for this is to have data dashboards for the stewarded areas to support targeted fixing / improving and to build a stronger local discussion and community around data > information > knowledge. On Fri, Dec 7, 2012 at 11:32 AM, Michal Migurski wrote: > my intent for this is to produce county-by-county views that GIS managers in > different jurisdictions can use to understand whether their data is worth -- Martijn van Exel http://oegeo.wordpress.com/ http://openstreetmap.us/ ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
GeoTIFF version of that data: http://mike.teczno.com/img/osm-users-imports-2012-09.tif On Dec 7, 2012, at 10:32 AM, Michal Migurski wrote: > That's great! > > I've been thinking about some of the same things, inspired by Dennis > Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 kilometer > squares, measuring OSM user count (post-import) vs. possibility for > improvement between 2007 and 2012 TIGER/Line data: > > http://mike.teczno.com/img/osm-users-imports-2012-09.png > - Blue means 2012 data would be an improvement > - Red/Yellow means there are active OSM users there > > I used an Albers equal-area projection to ensure correct length and density > regardless of latitude, and my intent for this is to produce county-by-county > views that GIS managers in different jurisdictions can use to understand > whether their data is worth important, and what OSM users they'd need to > contact in order to be successful. > > I have a slippy map version of this but the projection registration is off by > a few hundred meters; need to figure out why that is and fix it. > > -mike. > > On Dec 7, 2012, at 10:00 AM, Ian Villeda wrote: > >> >> Hey team, >> >> Over here at MapBox we were inspired by Martjin's great post on identifying >> TIGER deserts[1], so we're attempting to effectively identify TIGER deserts >> for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). >> The goal is to create a slippy map from zoom level 2 to 15(?) such that it's >> easy to find the places in need of the most love. >> >> We have a osmium-genereated postigs database with all ways tagged as >> highway=* and their version number. The plan to run the analysis against a >> 5x5 km grid, although we will probably play with the size of the gridcell. >> Another option might be to conduct the analysis and varying levels of >> granularity. >> >> We're still consiering different methods of analysis - the goal being to >> normalize way density to display better identify TIGER deserts. We'd would >> love your thoughts / input / reactions. A few ideas we've tossed around / >> are activley pursuing: >> >> - percent of version 1 and/or 2 ways in each gridcell >> - average number of version 1/2 ways per gridcell >> - using a threshold value(s) similiar to Martjin's approach >> - [your ideas here] >> >> Ultimately we'd like showcase the rendered results in site that would make >> editing TIGER deserts easy. For now though this is just a heads up and an >> invitation if you have any thoughts on how better to identify TIGER deserts. >> >> thanks! >> >> [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ >> -- >> ian villeda >> mapbox | developmentseed >> https://twitter.com/ian_villeda >> >> >> ___ >> Talk-us mailing list >> Talk-us@openstreetmap.org >> http://lists.openstreetmap.org/listinfo/talk-us > > > michal migurski- contact info and pgp key: > sf/cahttp://mike.teczno.com/contact.html > > > > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us > michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
Re: [Talk-us] identifying TIGER deserts
That's great! I've been thinking about some of the same things, inspired by Dennis Zielstra's talk at SOTM-US. I'm looking at road lengths for 1x1 kilometer squares, measuring OSM user count (post-import) vs. possibility for improvement between 2007 and 2012 TIGER/Line data: http://mike.teczno.com/img/osm-users-imports-2012-09.png - Blue means 2012 data would be an improvement - Red/Yellow means there are active OSM users there I used an Albers equal-area projection to ensure correct length and density regardless of latitude, and my intent for this is to produce county-by-county views that GIS managers in different jurisdictions can use to understand whether their data is worth important, and what OSM users they'd need to contact in order to be successful. I have a slippy map version of this but the projection registration is off by a few hundred meters; need to figure out why that is and fix it. -mike. On Dec 7, 2012, at 10:00 AM, Ian Villeda wrote: > > Hey team, > > Over here at MapBox we were inspired by Martjin's great post on identifying > TIGER deserts[1], so we're attempting to effectively identify TIGER deserts > for the rest of the continental U.S. (sorry Hawaii + Alaska, you're next). > The goal is to create a slippy map from zoom level 2 to 15(?) such that it's > easy to find the places in need of the most love. > > We have a osmium-genereated postigs database with all ways tagged as > highway=* and their version number. The plan to run the analysis against a > 5x5 km grid, although we will probably play with the size of the gridcell. > Another option might be to conduct the analysis and varying levels of > granularity. > > We're still consiering different methods of analysis - the goal being to > normalize way density to display better identify TIGER deserts. We'd would > love your thoughts / input / reactions. A few ideas we've tossed around / are > activley pursuing: > > - percent of version 1 and/or 2 ways in each gridcell > - average number of version 1/2 ways per gridcell > - using a threshold value(s) similiar to Martjin's approach > - [your ideas here] > > Ultimately we'd like showcase the rendered results in site that would make > editing TIGER deserts easy. For now though this is just a heads up and an > invitation if you have any thoughts on how better to identify TIGER deserts. > > thanks! > > [1]: http://oegeo.wordpress.com/2012/10/21/binders-full-of-tiger-deserts/ > -- > ian villeda > mapbox | developmentseed > https://twitter.com/ian_villeda > > > ___ > Talk-us mailing list > Talk-us@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-us michal migurski- contact info and pgp key: sf/cahttp://mike.teczno.com/contact.html ___ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us
