We are about to add the "Convex Hull" analysis (https://github.com/CartoDB/cartodb/issues/8883) and would need some information for the screen that appears after an analysis is finished.
cc: @andrewxhill
Why not have a "Group points into polygons" analysis?
Inside of that analysis, there would be a set of choices,
etc?
Big +1 for me. And we could have convex hull in the description so it's easier to find for some people
I'm ok with this idea, but I'm adding @xavijam and @dgaubert to the conversation.
Fine by me, it would be only a change in the add-analysis modal view.
Also fine by me, In camshaft-reference we will have one analysis for each type.
I can change the UI to support that. I love the idea :)
@javierarce here you go:
{Group points into polygons} has finished.
{The geometries in your table were updated.}
* {the_geom}
{This column was updated to polygons encompassing the input points.}
* {category}
{If chosen, this will be the category over which the polygons were calculated.}
{Group points into polygons used group {method_chosen} to create new polygons.}
{Read more in [PostGIS documentation](http://postgis.net/docs/manual-2.2/)}
We obviously need better documentation on this one, so I put the general PostGIS link as filler for now.
Actually, will there be other aggregations that occur with this analysis? Like average values of all points in a category or something like that?
I don't see anything like that in the camshaft node: https://github.com/CartoDB/camshaft/blob/master/lib/node/nodes/convex-hull.js#L20
UI Changes are designed.
For the time being I'd go with the proposed 4 algorithms. @dgaubert / @rochoa any chance we can work on those additions during this sprint?

Concave will require an additional parameter I believe.
On Mon, Aug 8, 2016 at 5:51 AM Sergio Álvarez Leiva <
[email protected]> wrote:
UI Changes are designed.
For the time being I'd go with the proposed 4 algorithms. @dgaubert
https://github.com/dgaubert / @rochoa https://github.com/rochoa any
chance we can work on those additions during this sprint?[image: 01 layers - 06-170 analysis - group points in polygons]
https://cloud.githubusercontent.com/assets/132149/17476051/60fc2502-5d5e-11e6-9a6e-bc9b167b642d.png—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/CartoDB/cartodb/issues/9326#issuecomment-238191019,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAWmU6jAqNe61u3l3QCOhLdEz6cejDV5ks5qdvwPgaJpZM4JdhO9
.
_Andrew W. Hill, Chief Science Officer_
www.carto.com | @andrewxhill https://twitter.com/andrewxhill
Carto—Predict through Location
201 Moore Street, Brooklyn, New York, 11206, USA
The only parameter that I can imagine is the Allow Holes.
@javierarce, can we add a radio button set with
"ALLOW HOLES" - () YES (X) NO
?
No, the target precision is the key one.
On Mon, Aug 8, 2016 at 1:00 PM Sergio Álvarez Leiva <
[email protected]> wrote:
The only parameter that I can imagine is the Allow Holes.
@javierarce https://github.com/javierarce, can we add a radio button
set with
"ALLOW HOLES" - () YES (X) NOhttp://postgis.net/docs/ST_ConcaveHull.html
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/CartoDB/cartodb/issues/9326#issuecomment-238300871,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAWmU8PfL94yipREUQLbkb6g8JTugKGgks5qd2CQgaJpZM4JdhO9
.
_Andrew W. Hill, Chief Science Officer_
www.carto.com | @andrewxhill https://twitter.com/andrewxhill
Carto—Predict through Location
201 Moore Street, Brooklyn, New York, 11206, USA
True thing, I missed that one.
We need a picker as well like the one with significance. @javierarce let me know if you want me to do the mockup
Yes, please, @saleiva, this is starting to get quite complex.
Also, pinging @dgaubert & @rochoa for this: https://github.com/CartoDB/cartodb/issues/9326#issuecomment-238191019
Here is the UI for concave hull and for Convex Hull / Bounding Circle / Bounding Box


Looks great! One suggestion I have is changing "Algorithm" to "Method".
And will the grey text which says "define reference and target layers" be changed? If so, I propose "define reference layer and method".
@javierarce how does that sound?
I'd say it looks fantastic, thanks!
@dgaubert/@rochoa: have you thought of a name for this group of analyses? Would group-points be OK? (or maybe group-points-into-polygons?)
Updated: Silly me, I've just realized that we'll be using the names of each different analysis 😅 . So I propose to use the following names (but feel free to suggest a better alternative):
bounding-boxbounding-circleconvex-hullconcave-hull@ohasselblad: we'll need a title for this group of analyses and small sentence to show users in the "add a new analysis" window. Could you take care of writing that, please?
Or would the description we have now ('Calculates the minimum geometry that encloses all geometries of your dataset.') still be valid?
@javierarce, here you go!
Group points into polygons
Aggregate points into polygons such as convex hulls or bounding boxes.

My two cents related to Concave Hull:
:star: [ input box ] :red_circle:
Because that's the practical meaning of that parameter.
Hey guys, can we unblock this?
+1 to Abel's comment as far as it fits in the UI.
What is the bounding box for a point? I mean, if I don't use a group by and my source is a points dataset should I still get polygons?
@javierarce I prepared a branch with the new analyses: https://github.com/CartoDB/camshaft/pull/175.
Can you start using that for development? If you find any problem let me know.
@rochoa, you should get a single polygon containning all your features
What if I have just one point? 😈
SELECT ST_AsText(ST_Envelope('POINT(1 3)'::geometry));
st_astext
------------
POINT(1 3)
That's why I was 😈 . Because we are supposed to return polygons, but we won't for that particular case, unless we cover that corner case.
we can also disable the analysis with just one point datasets :P
Hey, convex hull options will be available soon, is this ticket solved there @javierarce ?
Yup
Most helpful comment
@javierarce, here you go!