presidential-precinct-map-2020 VS sf

Here is a link explaining why some states are incomplete. The NYTimes map is by precinct and some states report results by county or township instead of precinct. Therefore, the data is incompatible with the map.

The Times specifies in their “about the data” Github page why the remaining data won’t be displayed, since those lazy fuckers didn’t go as far as the 20-something giganerd did to assemble the entire nation’s data.

I'm trying to use the NYT precinct-level election results for an analysis in R, but having difficulty loading the data. The github post for the data is located here. And the data can be downloaded here.

Before you assume things are impossible, I’d highly recommend you actually look at the data for yourself. This dataset has actual precinct by precinct outcomes and you can see that it is absolutely expected that some precincts will vote overwhelmingly (98+%) for one candidate. Of course, if you believe you’ve found the story of the century (and it would be) that there was a precinct that had 10,000+ plus votes that were 100% for one candidate, I’m all ears.

I posted a link, which was from the NYTime's graphic, about the data. This is the github link ("About Data") from the black box in the upper left corner of the page. https://www.nytimes.com/interactive/2021/upshot/2020-election-map.html?referringSource=articleShare

The data is here: https://github.com/TheUpshot/presidential-precinct-map-2020

Ah, yeah, they actually give all the details on their github here.

They released the underlying data that you can parse yourself if you wanted: https://github.com/TheUpshot/presidential-precinct-map-2020

sf

Before learning about geostatistics, do you feel comfortable working with and exploring geospatial data? If not, I'd highly recommend getting comfortable with the sf package in R. It's an implementation of the OpenGIS standard in R tidyverse. The OpenGIS standard defines specific data types and functions for geospatial data, which means that you can read e.g. PostGIS documentation and use the same functions in R.

Tools used: various packages in R (tidycensus, dplyr, ggplot2, sf)

R supports working with spatial data really well; you should look into the sf-package: https://r-spatial.github.io/sf/

A lot of the books are out if date because geospatial has been rewritten from the ground up to be dramatically improved. I would focus on the sf package: https://r-spatial.github.io/sf/. I would also find a PostGIS book which sf shares many functions and learn to the the database when appropriate.

Are you using the sf package? I envision running a loop that creates a vector feature for each cell of the raster grid, intersects that feature with the underlying shapefile, multiplies the area of each intersected portion by its value, and assigns the raster the mean of those values. Kind of a lot to set up, but I'm not a master at this so maybe someone else knows a more straightforward method

I found this: https://github.com/r-spatial/sf/issues/1595

Now that I explained the concept, I will show some R code using the sf library to achieve this. sf stands for simple features and it's a very nice library for working with geospatial data.