2f:T8f3,# Using Boston EMS crash data in R Below is a quick guide to loading the Boston Vision Zero crash data into R and doing some basic analysis. R is a statistical analysis software tool. If you've never used it before, you can [download it here](https://www.r-project.org/) and learn the basics [with the swirl package](https://swirlstats.com/students.html). You might also want to download [RStudio](https://posit.co/download/rstudio-desktop/), an integrated development environment for R. You can download the Boston crash data [from the city's data.boston.gov website](https://data.boston.gov/dataset/vision-zero-crash-records/resource/e4bfe397-6bfc-49c5-9367-c879fac7401d). For this example, we're downloading the data through the end of 2024. You'll get different results if you use a more recent dataset. ## Working with the data First, we'll load our CSV file from the Downloads folder into R: setwd('~/Downloads') boscrash <- read.csv('tmpnxi6jqlb.csv') Let's check the structure of the dataframe we just loaded: summary(boscrash) ### Formatting dates I can see that I have some geographic coordinates, street names, and a `mode_type` column that specifies whether the crash involved a bike, pedestrian, or motorist. There's also a `dispatch_ts` column that I'm going to reformat as an R datetime object: boscrash$dispatch_ts<-as.Date(boscrash$dispatch_ts) summary(boscrash$dispatch_ts) Min. 1st Qu. Median Mean 3rd Qu. Max. "2015-01-01" "2017-04-08" "2019-06-01" "2019-08-01" "2021-11-06" "2024-12-31" ## Finding crashes on a specific street My dataset has 37,893 rows, representing crashes going back to 2015. But I want to know how many crashes happened on Tremont Street in the South End. Tremont Street is relatively straight, so I can save it as a SpatialLines object with relatively few latitude and longitude points. This is a simple example, but if you want to investigate a curvier street, you can download latitude/longitude coordinates with tools like the [Overpass API](https://wiki.openstreetmap.org/wiki/Overpass_API) from OpenStreetMap. First, I'm going to go to Google Maps and right-click on the map to get latitude and longitude coordinates for points along the street: 30:T8c6,Now I'll use the `geosphere` package's `Dist2Line` function to calculate the distance between each crash and the closest point along the Tremont Street line, using latitude and longitude coordinates from each crash: library(geosphere) boscrash$d <- dist2Line(cbind(boscrash$long, boscrash$lat), tremontStreet) This will take R a while – it's a lot of math – so be patient. You'll get a warning message about coordinate reference systems but don't worry about it, for our purposes these distance calculations will be close enough. At this point, I'll select a random row for the data and check to see if the distances we just calculated are accurate: boscrash[12,] dispatch_ts mode_type location_type street xstreet1 xstreet2 x_cord y_cord 12 2015-05-13 mv Intersection NIGHTINGALE ST BERNARD ST 769596.3 2933074 lat long d.distance d.lon d.lat d.ID 12 42.29574 -71.08074 4853.23821 -71.08023 42.33943 1.00000 So this describes a crash at Nightingale Street and Bernard Street at longitude -71.081, latitude 42.296. Checking Google Maps, I can see that that's an intersection near Harambee Park in Dorchester, and it's about 4.84 kilometers from there to the intersection of Tremont Street and Mass. Ave. in the South End – the westernmost point on our `tremontStreet` line object. So the distance calculation (4853.24 meters) seems to have worked. Note that the `dist2Line` function created a new object that doesn't just include the distance, but also includes the closest latitude/longitude coordinates of the nearest point along the line (in the example above, the `d.lat` and `d.lng` coordinates describe the westernmost point of Tremont Street at Mass. Ave.). I'm only interested in the `d.distance` number - the distance, in meters, between each crash site and Tremont Street: boscrash$distance <- boscrash$d[,1] And then I'm going to select all the crashes that are less than 20 meters away from the center of Tremont: tremontCrashes <- subset(boscrash, distance < 20) And now let's plot those crashes: plot(tremontStreet) points(cbind(tremontCrashes$long, tremontCrashes$lat), col='blue', pch=20) 31:T4bd,## Summarizing the number of crashes by year and mode type Now I'd like to analyze our filtered dataset of crashes on Tremont Street to look at the number of crashes in each calendar year. First, I'll add a `year` column from the `dispatch_ts` data using the `lubridate` package: library(lubridate) tremontCrashes$year <- year(tremontCrashes$dispatch_ts) Now I can use the `dplyr` package to generate a summary of crashes by year and mode type: library(dplyr) grouped_data <- tremontCrashes %>% group_by(year, mode_type) %>% summarize(count = n()) print(grouped_data) # A tibble: 29 x 3 # Groups: year [10] year mode_type count 1 2015 bike 5 2 2015 mv 5 3 2015 ped 6 4 2016 bike 1 5 2016 mv 9 6 2016 ped 8 7 2017 bike 1 8 2017 mv 11 9 2017 ped 10 10 2018 bike 1 # … with 19 more rows. I can also use the `ggplot` package to draw a bar chart of the grouped data: library(ggplot2) ggplot(tremontCrashes, aes(x=year, fill=mode_type)) + + geom_bar(stat="count") 32:[{"id":"fa95e0ca93920e78835cd6e11418574858a310e24b379878a6f8d9999f2159fef1a6257bc870102af0b09807ff780201118427037a1668ec0ee4e3893588df85","href":"https://static.observableusercontent.com/files/fa95e0ca93920e78835cd6e11418574858a310e24b379878a6f8d9999f2159fef1a6257bc870102af0b09807ff780201118427037a1668ec0ee4e3893588df85","name":"Screenshot 2025-03-25 at 3.22.34 PM.png","type":"image/png","lastModified":1742930570391.182,"size":1112372},{"id":"46712897023fd1fba96c6e5d3ae48b97f56a84ba85d26ffb8c127775f5009616e296adac4860299c9a1514bd22a71b544ca2d9c2d012f57e0f3be296a216bcae","href":"https://static.observableusercontent.com/files/46712897023fd1fba96c6e5d3ae48b97f56a84ba85d26ffb8c127775f5009616e296adac4860299c9a1514bd22a71b544ca2d9c2d012f57e0f3be296a216bcae","name":"tremontPlot.png","type":"image/png","lastModified":1742931005043.779,"size":38090},{"id":"85d0bb59c4ddfe50f305c3fa43994b308b4678b8661e007ecd3abeaa3c46a5274495bc34cd8406d924acd3b835a7616b06d48a36b3355890abe9db0639575ebf","href":"https://static.observableusercontent.com/files/85d0bb59c4ddfe50f305c3fa43994b308b4678b8661e007ecd3abeaa3c46a5274495bc34cd8406d924acd3b835a7616b06d48a36b3355890abe9db0639575ebf","name":"TremontCrashesRPlot.png","type":"image/png","lastModified":1742932652757.05,"size":39575},{"id":"85d0bb59c4ddfe50f305c3fa43994b308b4678b8661e007ecd3abeaa3c46a5274495bc34cd8406d924acd3b835a7616b06d48a36b3355890abe9db0639575ebf","href":"https://static.observableusercontent.com/files/85d0bb59c4ddfe50f305c3fa43994b308b4678b8661e007ecd3abeaa3c46a5274495bc34cd8406d924acd3b835a7616b06d48a36b3355890abe9db0639575ebf","name":"TremontCrashesRPlot@1.png","type":"image/png","lastModified":1742932667191.745,"size":39575},{"id":"90967545bbb73144ca1629d0bdb795aba79ae37f28c4af5056c71aafbd324e3be9576250239020a229ce5e1e6dd27406414f3a4baf92053c59e7c53b87c3fedd","href":"https://static.observableusercontent.com/files/90967545bbb73144ca1629d0bdb795aba79ae37f28c4af5056c71aafbd324e3be9576250239020a229ce5e1e6dd27406414f3a4baf92053c59e7c53b87c3fedd","name":"TremontCrashBarPlot.png","type":"image/png","lastModified":1742933552460.535,"size":47521}] 2d:{"title":"Using Boston EMS crash data in R","cells":[{"id":0,"value":"$2f","pinned":false,"hidden":false,"mode":"md"},{"id":11,"value":"screenshot20250325At32234Pm = FileAttachment(\"Screenshot 2025-03-25 at 3.22.34 PM.png\").image()","pinned":true,"hidden":false,"mode":"ojs"},{"id":15,"value":"There's a slight bend in Tremont Street, but four sets of coordinates traces the centerline of the street very accurately. I'm going to create two vectors, one with the latitudes, and one with the longitudes:\n\n lng <- c(-71.080145, -71.07064, -71.07038, -71.06991, -71.06868)\n lat <- c(42.33944, 42.34451, 42.34468, 42.34520, 42.34738)\n\nNow I need to load some libraries for handling geographic data:\n\n library(raster)\n library(sp)\n library(geosphere)\n library(rgeos)\n\nAnd I'll create a new SpatialLines object from my coordinates:\n\n tremontStreet <- as(sp::SpatialPoints(cbind(lng, lat)), \"SpatialLines\")\n plot(tremontStreet)\n\nThe plot should look like this - a very basic map of Tremont Street in the South End:","pinned":false,"hidden":false,"mode":"md"},{"id":20,"value":"tremontplot = FileAttachment(\"tremontPlot.png\").image()","pinned":true,"hidden":false,"mode":"ojs"},{"id":23,"value":"$30","pinned":false,"hidden":false,"mode":"md"},{"id":30,"value":"tremontcrashesrplot1 = FileAttachment(\"TremontCrashesRPlot@1.png\").image()","pinned":true,"hidden":false,"mode":"ojs"},{"id":33,"value":"$31","pinned":false,"hidden":false,"mode":"md"},{"id":38,"value":"tremontcrashbarplot = FileAttachment(\"TremontCrashBarPlot.png\").image()","pinned":true,"hidden":false,"mode":"ojs"}],"files":"$W32","stdlib":"1","titling":"auto"}