4 ms·

Why is there such a movement for making maps with R? The python ecosystem has more and better packages for making maps already. Notwithstanding that there are i

Why is there such a movement for making maps with R? The python ecosystem has more and better packages for making maps already. Notwithstanding that there are indeed some nice results, this new R toolchain movement towards maps feels like a waste of open source energy to me.

jackdaniel 3y ago
Can we, I don't know, let people enjoy programming languages they like without calling it a waste?
minimaxir 3y ago
As someone proficient with R and Python, anything data viz-wise is substantially easier to do in R/ggplot2, particularly something complicated like this post. Most of the R packages in this post aren't that new either.
- pjacotg 3y ago
  I agree with that. I've been programming on and off in Python since 2008 and really like the language. I started using R just over a year ago after joining a new team. It took a while to adjust but I've gotten to prefer it for exploratory data analysis and data viz.
- jarenmf 3y ago
  My same experience, I have built fairly big projects with Python and I like it for general tasks but whenever I have something data analytics/visualization related I find myself reaching for R. There is so much functionality built into the language that just makes me so efficient.
- dacryn 3y ago
  same here, I love me some ggplot. Python for everything, but if I need to make something nice to look at, R makes my life so much easier, and the final touch in Affinity Designer/Photo. In jupyter it's trivial to mix and match python and R anyway, all the wrangling in python and once it's clean, just basic tidyverse in R
a_bonobo 3y ago
I feel like the R ecosystem was there first; but perhaps two completely separate groups of people. Ecologists, researchers working with geological or geospatial data use R and have been using R for a long, long time. The 'old 'standard library 'maps' is decades old by now, with maps v3 now being 8 years old. maps v2 goes back to 2003, about the time of Python 2.3? I can't even find the sources of maps v1. https://cran.r-project.org/src/contrib/Archive/maps/ https://cran.r-project.org/src/contrib/Archive/maps/ I believe that maps v1 was for R's predecessor, S, so from the late 90s, about the time Python 1.5 came out. Was anybody even using Python then? OP's post uses the slightly fancier tidyverse-versions of these decades-old libraries. I think it's safe to say that map-making libraries are older in R than in Python.
ezconnect 3y ago
There are a lot of R users and it's a nice language you should try it sometimes.
- _Wintermute 3y ago
  As an R user for nearly 15 years at this point: R isn't nice, it's an awful language, but it is very useful.
  - doctaj 3y ago
    The tidyverse has made it pretty “nice”.
anaganisk 3y ago
Why build react, angular and all, when jQuery was just fine, why is there such a moment in Js to always build the next framework, seems like a waste of open source energy. ....something....something...requirement.. different approach...
m2fkxy 3y ago
The mapmaking R ecosystem absolutely has been around for ages, and from a cartographer's point of view it is better suited at making actual maps than Python.
- smcl 3y ago
  As a non-cartographer, but someone who loves maps (who doesnt?) and programming is there anything in particular that makes R more suitable than Python? Not trying to start a Python/R flame war, genuinely curious as an outsider if there's some secret sauce that R has but which Python lacks.
  - m2fkxy 3y ago
    IMO it mostly boils down to good practices in symbology, projection, legend, scaling, etc. being better implemented and easy-to-use for the map maker. ggplot2 is an obvious heavy-weight, but other packages such as mapsf and tmap also are top notch in this regard. on the other hand, my general map making experience in Python has not been very pleasant when I had to work towards creating a map following acceptable cartographic practices, ie. anything beyond throwing data on a webmap layer with subpar legend and symbology.
  - j7ake 3y ago
    The secret sauce for R is that their statistical models/packages are cutting-edge, they have the best plotting library (ggplot2), and they have the best data wrangling packages (dplyr for tabular data).
    - countrymile 3y ago
      very much this. I've spent years dabbling with other languages, nothing as ever been as clean as the tidyverse for data analysis
b86754 3y ago
Interesting, I’ve used both for years and have found R to be way more intuitive for data exploration and visualization. R looks different than most programming languages, but once it clicks it’s actually really nice.
- queens_ridge 3y ago
  ggplot is the best! R has some fantastic packages especially for visualizations. And the community is great :)
j7ake 3y ago
You're welcome to demonstrate why python ecosystem would be better for this analysis, otherwise you're making unsubstantiated claims.
nologic01 3y ago
while I use Python far more than R (or Julia) I don't think that this is a particularly constructive comment. "More and better" is simply subjective and not substantiated. In fact detailed and objective comparisons of capabilities, performance etc between these ecosystems are sorely lacking not just in map-making but practially concerning every comparison. Still, the "waste of open source energy argument" is quite bit more complex and I'm frequently wondering about this. For sure this "waste" happens everywhere in open source. About the only thing that approaches some optimal level of efficiency is the Linux kernel. From there on, distributions, desktops, toolchains, apps, libraries etc. is the epitomy of chaos. This diversity creates resilience and a sort of competition for the survival of the fittest. The openness also facilitates fast diffusion and borrowing good ideas. But it is also true that open source resources are still scarce so development happens at maybe much lower speed than what would be possible if more coordination / less fragmentation was practiced.

jackdaniel 3y ago

Can we, I don't know, let people enjoy programming languages they like without calling it a waste?

minimaxir 3y ago

As someone proficient with R and Python, anything data viz-wise is substantially easier to do in R/ggplot2, particularly something complicated like this post. Most of the R packages in this post aren't that new either.

pjacotg 3y ago
I agree with that. I've been programming on and off in Python since 2008 and really like the language. I started using R just over a year ago after joining a new team. It took a while to adjust but I've gotten to prefer it for exploratory data analysis and data viz.
jarenmf 3y ago
My same experience, I have built fairly big projects with Python and I like it for general tasks but whenever I have something data analytics/visualization related I find myself reaching for R. There is so much functionality built into the language that just makes me so efficient.
dacryn 3y ago
same here, I love me some ggplot. Python for everything, but if I need to make something nice to look at, R makes my life so much easier, and the final touch in Affinity Designer/Photo. In jupyter it's trivial to mix and match python and R anyway, all the wrangling in python and once it's clean, just basic tidyverse in R

a_bonobo 3y ago

I feel like the R ecosystem was there first; but perhaps two completely separate groups of people. Ecologists, researchers working with geological or geospatial data use R and have been using R for a long, long time. The 'old 'standard library 'maps' is decades old by now, with maps v3 now being 8 years old. maps v2 goes back to 2003, about the time of Python 2.3? I can't even find the sources of maps v1. https://cran.r-project.org/src/contrib/Archive/maps/ https://cran.r-project.org/src/contrib/Archive/maps/ I believe that maps v1 was for R's predecessor, S, so from the late 90s, about the time Python 1.5 came out. Was anybody even using Python then? OP's post uses the slightly fancier tidyverse-versions of these decades-old libraries. I think it's safe to say that map-making libraries are older in R than in Python.

ezconnect 3y ago

There are a lot of R users and it's a nice language you should try it sometimes.

_Wintermute 3y ago
As an R user for nearly 15 years at this point: R isn't nice, it's an awful language, but it is very useful.
- doctaj 3y ago
  The tidyverse has made it pretty “nice”.

anaganisk 3y ago

Why build react, angular and all, when jQuery was just fine, why is there such a moment in Js to always build the next framework, seems like a waste of open source energy. ....something....something...requirement.. different approach...

m2fkxy 3y ago

The mapmaking R ecosystem absolutely has been around for ages, and from a cartographer's point of view it is better suited at making actual maps than Python.

smcl 3y ago
As a non-cartographer, but someone who loves maps (who doesnt?) and programming is there anything in particular that makes R more suitable than Python? Not trying to start a Python/R flame war, genuinely curious as an outsider if there's some secret sauce that R has but which Python lacks.
- m2fkxy 3y ago
  IMO it mostly boils down to good practices in symbology, projection, legend, scaling, etc. being better implemented and easy-to-use for the map maker. ggplot2 is an obvious heavy-weight, but other packages such as mapsf and tmap also are top notch in this regard. on the other hand, my general map making experience in Python has not been very pleasant when I had to work towards creating a map following acceptable cartographic practices, ie. anything beyond throwing data on a webmap layer with subpar legend and symbology.
- j7ake 3y ago
  The secret sauce for R is that their statistical models/packages are cutting-edge, they have the best plotting library (ggplot2), and they have the best data wrangling packages (dplyr for tabular data).
  - countrymile 3y ago
    very much this. I've spent years dabbling with other languages, nothing as ever been as clean as the tidyverse for data analysis

b86754 3y ago

Interesting, I’ve used both for years and have found R to be way more intuitive for data exploration and visualization. R looks different than most programming languages, but once it clicks it’s actually really nice.

queens_ridge 3y ago
ggplot is the best! R has some fantastic packages especially for visualizations. And the community is great :)

j7ake 3y ago

You're welcome to demonstrate why python ecosystem would be better for this analysis, otherwise you're making unsubstantiated claims.

nologic01 3y ago

while I use Python far more than R (or Julia) I don't think that this is a particularly constructive comment. "More and better" is simply subjective and not substantiated. In fact detailed and objective comparisons of capabilities, performance etc between these ecosystems are sorely lacking not just in map-making but practially concerning every comparison. Still, the "waste of open source energy argument" is quite bit more complex and I'm frequently wondering about this. For sure this "waste" happens everywhere in open source. About the only thing that approaches some optimal level of efficiency is the Linux kernel. From there on, distributions, desktops, toolchains, apps, libraries etc. is the epitomy of chaos. This diversity creates resilience and a sort of competition for the survival of the fittest. The openness also facilitates fast diffusion and borrowing good ideas. But it is also true that open source resources are still scarce so development happens at maybe much lower speed than what would be possible if more coordination / less fragmentation was practiced.