by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jan 31 2019 21:43
    davidbrochart starred pysal/pysal
  • Jan 31 2019 18:24
    sercanerhan starred pysal/pysal
  • Jan 31 2019 17:48
    klausvesama starred pysal/pysal
  • Jan 31 2019 15:51
    Tafkas starred pysal/pysal
  • Jan 31 2019 15:12
    mxfh starred pysal/pysal
  • Jan 31 2019 14:49
    sjsrey starred pysal/pysal
  • Jan 31 2019 11:10
    unverciftci starred pysal/pysal
  • Jan 31 2019 03:28
    jcbain starred pysal/pysal
  • Jan 31 2019 03:25
    ifarah starred pysal/pysal
  • Jan 30 2019 20:17

    sjsrey on master

    configure readthedocs remove .idea Merge pull request #1079 from w… (compare)

  • Jan 30 2019 20:17
    sjsrey closed #1079
  • Jan 30 2019 20:17
    weikang9009 review_requested #1079
  • Jan 30 2019 20:16
    weikang9009 opened #1079
  • Jan 30 2019 20:08

    weikang9009 on master

    DOC: readthedocs files Merge pull request #1078 from s… (compare)

  • Jan 30 2019 20:08
    weikang9009 closed #1078
  • Jan 30 2019 20:01
    sjsrey opened #1078
  • Jan 30 2019 20:01
    sjsrey assigned #1078
  • Jan 30 2019 19:31

    sjsrey on v2.0.0

    (compare)

  • Jan 30 2019 19:28

    sjsrey on master

    DOC: Installation of legacy DOC: installation DOC: Update changelog and 1 more (compare)

  • Jan 30 2019 19:28
    sjsrey closed #1077
Francisco Ferioli Marco
@franciscoferiolimarco_gitlab
Good morning, sorry for bothering, I hope I'm not breaking any rule. I'm Francisco from Argentina. I wrote an email to Stefanie Lumnitz a month ago in which I asked her some things and proposed others in relationship to her project aiTree.
Francisco Ferioli Marco
@franciscoferiolimarco_gitlab
This is the first time I find other method for contacting her, appart from her email. I hope she reads this, or alternatively, if you don't mind, please tell her I want to talk with her, it's really important for me. I want to map as many trees as possible and her library is 1 of the only 3 existing open source projects of its kind in the whole world, as far as I know, that without mentioning that the second project, Treepedia, has some important problems which prevent me from using it.
Have a nice week. Greetings.
Stefanie Lumnitz
@slumnitz
Hello @franciscoferiolimarco_gitlab nice to e-meet you! Thank you for reaching out via gitter, I’ll make sure to respond to you e-mails shortly!
Dani Arribas-Bel
@darribas
I'm ready for the W ids discussion, do we have a room?
Stefanie Lumnitz
@slumnitz
not yet, shall we go for a google meet or is someone with a fancy zoom around? :)
Sergio Rey
@sjsrey
for the weights meeting here is the room: To join the video meeting, click this link: https://meet.google.com/yif-ujjj-ykh
Levi John Wolf
@ljwolf
hi sorry missed this
please let me in when possible.
Francisco Ferioli Marco
@franciscoferiolimarco_gitlab
Hello Stefanie, thank you for your answer.
Levi John Wolf
@ljwolf
a bit of good news: our weights constructor is still faster than PyGEOS using an STRtree (270s vs . 164s. for all the LSOAs in scotland, england, & wales, n=41729). But, that comes at a cost of twice the memory pressure.
eli knaap
@knaaptime
go speed racer
Levi John Wolf
@ljwolf
he's very heavy :)
Martin Fleischmann
@martinfleis
which function is that?
Levi John Wolf
@ljwolf
given that sets are now supported in numba, we might see a further speed/mem benefit to exploring numba-fying the algo
@martinfleis libpysal.weights.Queen.from_dataframe(), which relies on the contiguity builder within libpysal.weights._contw_lists
Martin Fleischmann
@martinfleis
oh yeah. But that does not do the same thing as pygeos, i.e. you need proper topology. pygeos gives you neighbour even if it does not share vertex, so it is a bit unfair :)
Levi John Wolf
@ljwolf
the equivalent pygeos op would be:
pygeos_pgons = pygeos.from_shapely(lsoas.geometry)
rtree = pygeos.STRtree(pygeos_pgons)
all_pairs_touches = rtree.query_bulk(pygeos_pgons, predicate='touches')
yeah, of course
I think this old comparison is our target
for lattice data, the vertices are (by definition) shared, so that's generally the comparison we're focused on... but when you get those non-planarities, our stuff breaks down
Martin Fleischmann
@martinfleis
this reminds me that we can use query_bulk in fuzzy_contiguity if the user has geopandas 0.8 to significantly speed that up.
Levi John Wolf
@ljwolf
ah, yes, great thinking!
Levi John Wolf
@ljwolf
Hey folks, the summer of code call is here today: https://zoom.us/j/96200473684?pwd=QWJDZktZeWtpRlkxbkwreHZQakl4Zz09
Levi John Wolf
@ljwolf
real interesting work from Chris & Martin on local correlation contributions
James Gaboardi
@jGaboardi
Where is this originally from/published?
that might be a more useful way of doing some of our tests
Dani Arribas-Bel
@darribas
yo, folks up for a social drink after the GSoC calls today?
for post-meeting social
Dani Arribas-Bel
@darribas
habemus release?
Sergio Rey
@sjsrey
Two things left: [1] need feedback on pysal/pysal#1189 and [2] someone needs to push pointpats on pypi
Dani Arribas-Bel
@darribas
whoops! had missed that PR...
Sergio Rey
@sjsrey
James Gaboardi
@jGaboardi
:confetti_ball:
Stefanie Lumnitz
@slumnitz
🚀
Levi John Wolf
@ljwolf
woohoo
Levi John Wolf
@ljwolf
hello GSOC students! please attend the call today at this link: https://zoom.us/j/99477034536?pwd=Nk1WV3I3WHJTeFV6K3NwZ1J5ZnFMdz09
wendellwt
@wendellwt
Hello pysal! This project looks interesting and I'm wondering how I can use it.
wendellwt
@wendellwt
I have a geopandas gdf with a column of linestrings, and another geopandas gdf with a column of polygons. I would like to make some sort of "cross-product intersection" and get out a geopandas gdf with two geometry columns, one column consisting of the multilinestring (or geomtrycollection) of where each linestring from the first df intersected any polygon from the second; and the the other geometry column of the 'difference' (i.e., where the linestring did not intersect any of the polygons). Does pysal do something like this?
6 replies
Martin Fleischmann
@martinfleis
Just to clarify, geopandas.overlay works on mixed geometry types. You can control the behaviour using keep_geom_type argument.
wendellwt
@wendellwt
Ok, thanks. It seems that the keep_geom_type requires 0.8.1 which requires pygeos. I'll install those and try it.
Dmitry
@dementiev
Hi all. I have one "dependent" layer and many "predictor" layers (all of them are shapefiles with multipolygons). I want to run an analysis to say which predictor layers influence the dependent layer most. Another question is how is better to visualize it? I thought about a simple bar chart, but maybe there are better visualizations.
Sergio Rey
@sjsrey
Reminder about Friday's dev meeting. Draft agenda.