Comments on: Choropleths in R (yes, "choropleths") http://www.thisisthegreenroom.com/2009/choropleths-in-r/ in stereo where available Sun, 05 Feb 2012 19:56:05 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Damian http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-9294 Damian Fri, 11 Mar 2011 01:28:33 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-9294 I'm having trouble with this line, as well. It returns the following error: Error in `[.default`(xj, i) : invalid subscript type 'list' Any tips? I'm having trouble with this line, as well. It returns the following error:
Error in `[.default`(xj, i) : invalid subscript type 'list'

Any tips?

]]> By: Quora http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-9047 Quora Fri, 04 Mar 2011 06:29:43 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-9047 <strong>What is the easiest way to make a state-by-state choropleth map from data in a .csv?...</strong> I'd be interested in using whatever method does this the easiest. Ideally, I'd like to modify an SVG somehow and then import it into illustrator to pretty it up, but I'd be content with using something like polymaps to do so. I am struggling to find... What is the easiest way to make a state-by-state choropleth map from data in a .csv?...

I'd be interested in using whatever method does this the easiest. Ideally, I'd like to modify an SVG somehow and then import it into illustrator to pretty it up, but I'd be content with using something like polymaps to do so. I am struggling to find...

]]> By: Miriam http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-8733 Miriam Wed, 23 Feb 2011 05:50:16 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-8733 Sorry! For some reason, part of my post was chopped. The line I am stumbling over is: "states <- as.character(stateAbbr[states,2])" from the previous line I obtain for "states" something like this (this is the first entry): 'states[1] Autauga County, AL “AL”' What does "stateAbbr[states,2]" do, then? Thank you so much! Sorry! For some reason, part of my post was chopped.
The line I am stumbling over is:

"states <- as.character(stateAbbr[states,2])"

from the previous line I obtain for "states" something like this (this is the first entry):

'states[1]
Autauga County, AL
“AL”'

What does "stateAbbr[states,2]" do, then?

Thank you so much!

]]> By: Miriam http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-8732 Miriam Wed, 23 Feb 2011 05:45:34 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-8732 Hello! I am trying to replicate your awesome graph, and I think I have almost everything working except this one line: 'states states[1] Autauga County, AL "AL" ) Thank You! Hello!

I am trying to replicate your awesome graph, and I think I have almost everything working except this one line:

'states states[1]
Autauga County, AL
"AL"
)

Thank You!

]]> By: thomas http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-8648 thomas Sat, 19 Feb 2011 14:01:28 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-8648 found it. came from me forgetting to order the code, this line: mergedata <- mergedata[order(mergedata$order),] although i don't understand why not running it would lead to white gaps? found it. came from me forgetting to order the code, this line:
mergedata <- mergedata[order(mergedata$order),]

although i don't understand why not running it would lead to white gaps?

]]> By: thomas http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-8631 thomas Sat, 19 Feb 2011 01:16:11 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-8631 hi, i tried your code with my own data, but only on state level. Code works well and is truly easy but for whatever reason, some white area is visible in certain states. if you look at california .... http://img39.imageshack.us/i/65881540.jpg/ any idea what causes that? appreciated ;) hi,
i tried your code with my own data, but only on state level. Code works well and is truly easy but for whatever reason, some white area is visible in certain states. if you look at california .... http://img39.imageshack.us/i/65881540.jpg/
any idea what causes that? appreciated ;)

]]> By: J http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-3366 J Sun, 03 Oct 2010 15:56:48 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-3366 Hi Jon -- I'm pretty sure there is a way to export the contours of the map as a series of lat/long coordinates (though I can't say off the bat how). You could pair that data with a z-variable to create the 3D graph. I would be remiss if I didn't caution that 3D charts are frequently more about eye-candy and less about information; make sure that the move to 3D actually imparts greater meaning in addition to looking cool. Please let us know if it works! Hi Jon -- I'm pretty sure there is a way to export the contours of the map as a series of lat/long coordinates (though I can't say off the bat how). You could pair that data with a z-variable to create the 3D graph. I would be remiss if I didn't caution that 3D charts are frequently more about eye-candy and less about information; make sure that the move to 3D actually imparts greater meaning in addition to looking cool.

Please let us know if it works!

]]> By: jonmcrawford http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-3351 jonmcrawford Sat, 02 Oct 2010 12:29:53 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-3351 I was wondering if you could start with this type of data, then transform the plots into the persp() package to create a prism map? (intensity would not be just by color, but then elevation of the county in the image, with a consistent z-value for the entire county) Not sure if that would work, but I was going to play around with that. I was wondering if you could start with this type of data, then transform the plots into the persp() package to create a prism map? (intensity would not be just by color, but then elevation of the county in the image, with a consistent z-value for the entire county)

Not sure if that would work, but I was going to play around with that.

]]> By: Matthew http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-2675 Matthew Fri, 16 Jul 2010 07:20:26 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-2675 Hi, With respect to adding Hawaii and Alaska: Why not transform the latittude and and longitude values describing their boundaries so they are situated south of the mainland states as per Nathan's map? Nice work by the way. Cheers Matt Legge Hi,
With respect to adding Hawaii and Alaska: Why not transform the latittude and and longitude values describing their boundaries so they are situated south of the mainland states as per Nathan's map?
Nice work by the way.
Cheers

Matt Legge

]]> By: J http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-2560 J Wed, 09 Jun 2010 13:53:54 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-2560 Hi Clem - Sorry for the confusion. I guess my code isn't quite as self-contained as I had thought. "state abbr.csv" is a simple file I created to provide a mapping between state names and state abbreviations -- as I recall, the data format and the necessary format in R were not the same. The file looked like this: New Jersey | NJ || New Mexico | NM || New York | NY || ... etc. A very simple translation from one set of data into another. "unemp_data" is my [unemp]loyment [data]set. To have R do any useful plotting, the plotted object needs to contain all plotted attributes -- it's very hard to overlay multiple objects intelligently. So, my main dataset needs to contain not just the data to be plotted, but extraneous items like the colors I want to plot in as well (note there are ways around this if you are willing to let R pick colors on your behalf). In this specific line, I am using the cut function to define "breaks" in the unemployment data of 2% width, capped at 100% to fill out the last bucket. So a level of 3% would be in one bucket and a level of 6% would be in another bucket. I convert those buckets to numbers (again, I don't remember exactly why but most likely just as a data integrity step) and then to factors, which tells R to consider each value a distinct "level". In other words, it won't think of 1,2,3,4 as a continuous spectrum in which "2" is near "3" but far from "4"; rather, it will think of each number as a "level" and there is no relationship between "2" and "3" that implies greater similarity than between "2" and "4". Finally, I store my color factors in the dataset's colorBuckets column. Later, I tell the plotting functions to give each of those buckets a different color, which I also specify. Hi Clem -
Sorry for the confusion. I guess my code isn't quite as self-contained as I had thought.

"state abbr.csv" is a simple file I created to provide a mapping between state names and state abbreviations -- as I recall, the data format and the necessary format in R were not the same. The file looked like this: New Jersey | NJ || New Mexico | NM || New York | NY || ... etc. A very simple translation from one set of data into another.

"unemp_data" is my [unemp]loyment [data]set. To have R do any useful plotting, the plotted object needs to contain all plotted attributes -- it's very hard to overlay multiple objects intelligently. So, my main dataset needs to contain not just the data to be plotted, but extraneous items like the colors I want to plot in as well (note there are ways around this if you are willing to let R pick colors on your behalf). In this specific line, I am using the cut function to define "breaks" in the unemployment data of 2% width, capped at 100% to fill out the last bucket. So a level of 3% would be in one bucket and a level of 6% would be in another bucket. I convert those buckets to numbers (again, I don't remember exactly why but most likely just as a data integrity step) and then to factors, which tells R to consider each value a distinct "level". In other words, it won't think of 1,2,3,4 as a continuous spectrum in which "2" is near "3" but far from "4"; rather, it will think of each number as a "level" and there is no relationship between "2" and "3" that implies greater similarity than between "2" and "4". Finally, I store my color factors in the dataset's colorBuckets column. Later, I tell the plotting functions to give each of those buckets a different color, which I also specify.

]]> By: Clem http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-2552 Clem Tue, 08 Jun 2010 12:23:46 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-2552 Hi, I am trying to use your code to do a similar map of the US by county of the mortality rate caused by colon cancer. I am not a pro at R therefore i have a few question about your code. how or where did you get the file "state abbr.csv" ? And could briefly explain this line of code not sure i understand what "unemp" is unemp_data$colorBuckets <- as.factor(as.numeric(cut(unemp_data$unemp,c(0,2,4,6,8,10,100)))) Thank you and great post extremely useful Hi,
I am trying to use your code to do a similar map of the US by county of the mortality rate caused by colon cancer. I am not a pro at R therefore i have a few question about your code.
how or where did you get the file "state abbr.csv" ?
And could briefly explain this line of code not sure i understand what "unemp" is

unemp_data$colorBuckets <- as.factor(as.numeric(cut(unemp_data$unemp,c(0,2,4,6,8,10,100))))

Thank you and great post extremely useful

]]> By: iPhone Simulator causes package installation to hang on OS X - As Yet Unclassified http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-2390 iPhone Simulator causes package installation to hang on OS X - As Yet Unclassified Thu, 18 Mar 2010 22:51:54 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-2390 [...] programming language. I’d heard of it in passing, but hadn’t tried it. After seeing how cool maps could be generated with very little code, I decided to try it [...] [...] programming language. I’d heard of it in passing, but hadn’t tried it. After seeing how cool maps could be generated with very little code, I decided to try it [...]

]]> By: Joe http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-1622 Joe Sat, 14 Nov 2009 13:20:01 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-1622 I tried doing this a few months ago in python, got lazy and found this http://www.sacmeq.org/statplanet/ Its all done in excel which makes it easy. Gives you a time scale on the bottom too. I tried doing this a few months ago in python, got lazy and found this
http://www.sacmeq.org/statplanet/
Its all done in excel which makes it easy. Gives you a time scale on the bottom too.

]]> By: J http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-1615 J Fri, 13 Nov 2009 05:23:57 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-1615 Hadley - I was about to say thanks, because through you I discovered the embeddable code templates that I used here, but I just realized that I owe you much more because ggplot2 is actually <a href="http://had.co.nz/" rel="nofollow">yours</a>! Knowing that, I'm not sure a simple thanks would do - it really is a phenomenal package (and I think I'm going to become its newest evangelist). Hadley - I was about to say thanks, because through you I discovered the embeddable code templates that I used here, but I just realized that I owe you much more because ggplot2 is actually yours! Knowing that, I'm not sure a simple thanks would do - it really is a phenomenal package (and I think I'm going to become its newest evangelist).

]]> By: Hadley http://www.thisisthegreenroom.com/2009/choropleths-in-r/comment-page-1/#comment-1614 Hadley Fri, 13 Nov 2009 04:21:37 +0000 http://www.thisisthegreenroom.com/?p=3123#comment-1614 I think the maps package does some thinning (generalisation) of the data which results in a slightly nicer appearance. I think if you look at the tip of Michigan, you can see a small difference between the state and county boundaries when you do this. I have some draft code to do this for general polygons, but it's very hard to do it in a topologically consistent manner. Nice plots :) I think the maps package does some thinning (generalisation) of the data which results in a slightly nicer appearance. I think if you look at the tip of Michigan, you can see a small difference between the state and county boundaries when you do this. I have some draft code to do this for general polygons, but it's very hard to do it in a topologically consistent manner.

Nice plots :)

]]>