jebyrnes / state_leg Goto Github PK

To use these scripts, you will need R installed. I also recomment using R Studio to run it. Once you install R, you'll also need a few packages. For baseline data entry, in R, run

install.packages("tidyverse")
install.packages("readxl")
install.packages("stringr")

If you want to use the map maping scripts:

install.packages("sp")
install.packages("rgdal")
install.packages("ggplot2")
install.packages("leaflet")
install.packages("forcats")

To get data in,

1. Find the appropriate elections data tables from Ballotpedia from pages such as https://ballotpedia.org/Alaska_House_of_Representatives_elections,_2016 - it will look like

1. Copy this table and paste it into excel. Make sure the top line of the spreadsheet is "District, Democrat, Republican, Other". Save into the folder raw_data as an excel file (not a CSV!) and closeit .

2. Once you have all of the tables you want, open R/01_make_clean_votes.R. Make sure your working directory is set to the directory that file is in. E.g., in Rstudio, go to the Session Menu, then select Set Working Director > To Source File Location.

3. Run the file. It will scroll through all of the files in the raw_data folder and spit out nicely formatted data into the clean_data folder.

That's it!

This is still in progress, but, you can visit https://www.census.gov/geo/maps-data/data/cbf/cbf_sld.html to get shapefiles of the 2015 state upper and lower legislative districts. Unzip the relevant shapefile into the spatial_data folder. I've provided Alaska as an example.

You then can open mapping_functions.R run it, and then you're free to make maps. For example, for the Alaska State House

ak_upper_combined <- make_district_data(shapefile = "cb_2015_02_sldu_500k",
                                         layer_name = "cb_2015_02_sldu_500k",
                                         votefile = "alaska_senate_2016.csv",
                                         spatial_dir = "./spatial_data/",
                                         vote_dir = "./clean_data/")

## OGR data source with driver: ESRI Shapefile 
## Source: "./spatial_data/cb_2015_02_sldu_500k", layer: "cb_2015_02_sldu_500k"
## with 20 features
## It has 9 fields

## Regions defined for each Polygons

## Joining, by = "id"

## Joining, by = "District"

## Warning: Unknown levels in `f`: Independent

This creates a dataset of district borders and information about each. I'll work up something for the sp library in the future so we can use leaflet. The data columns contained for plotting are

names(ak_upper_combined)

##  [1] "long"       "lat"        "order"      "hole"       "piece"     
##  [6] "group"      "id"         "STATEFP"    "SLDUST"     "AFFGEOID"  
## [11] "GEOID"      "District"   "LSAD"       "LSY"        "ALAND"     
## [16] "AWATER"     "State"      "Body"       "Year"       "Term"      
## [21] "Party"      "Prvote"     "Propvote"   "PrMargin"   "ThreePVote"

Note that for districts where there was no election that year, there are NAs for many values. make_district_data will put "No Election" in the Party column for these districts.

After this, you can make some nice maps, such as the party winners

#Make a nice map using ggplot2
ggplot(ak_upper_combined, mapping=aes(x=long, y=lat, group=group, fill=Party)) +
  geom_path() +
   geom_polygon() +
   xlim(c(-180, -130)) +
   theme_bw() +
   theme(panel.grid = element_blank()) +
   scale_fill_manual(values = c("Blue", "Red", "White"))

Or others (Happy to take suggestions here!)