There are yearly data about crime in Chicago at (1). On January 28, 2023 I downloaded the data Crimes_-_One_year_prior_to_present.csv from (2). Recently the map of wards changed.
A very general approach to producing a multiway network is to take a data frame with several nominal (or ordinal) variables and select the ways among them. Afterward, we transform it into the corresponding multiway network. For this task, I wrote the function DF2MWN.
> DF2MWN <- function(DF,ways,w=NULL,network="test",title="Test"){
+ info <- list(network=network,title=title,by="DF2MWN",date=date())
+ Ways <- as.list(ways); names(Ways) <- ways
+ data <- list()
+ nodes <- NULL; links <- data.frame(one=rep(1,nrow(CH)))
+ for(way in ways){
+ q <- factor(DF[[way]])
+ nodes[[way]] <- data.frame(ID=levels(q))
+ links[[way]] <- as.integer(q)
+ }
+ MN <- list(format="MWnets",info=info,ways=Ways,nodes=nodes,links=links,data=data)
+ return(MN)
+ }
The function DF2MWN is available in the library MWnets.
> setwd("C:/Users/vlado/DL/data/multi/chicago")
> library(jsonlite)
> library(Polychrome)
> library(magrittr)
> source("https://raw.githubusercontent.com/bavla/ibm3m/master/multiway/MWnets.R")
> CH <- read.table("Crimes_-_One_year_prior_to_present.csv",sep=",",header=FALSE,skip=1,fill=TRUE)
> dim(CH)
[1] 239616 17
> names(CH) <- c("case","date","block","iucr","primary","secondary","location","arrest","domestic",
+ "beat","ward","fbi","x","y","lat","lon","loc")
> head(CH)
> MT <- DF2MWN(CH,c("primary","location","arrest","domestic","ward","fbi"),
+ network="ChicagoCrime22",title="City of Chicago incidents of crime 2022")
> str(MT)
List of 6
$ format: chr "MWnets"
$ info :List of 4
..$ network: chr "ChicagoCrime22"
..$ title : chr "City of Chicago incidents of crime 2022"
..$ by : chr "DF2MWN"
..$ date : chr "Mon Jan 30 04:02:07 2023"
$ ways :List of 6
..$ primary : chr "primary"
..$ location: chr "location"
..$ arrest : chr "arrest"
..$ domestic: chr "domestic"
..$ ward : chr "ward"
..$ fbi : chr "fbi"
$ nodes :List of 6
..$ primary :'data.frame': 31 obs. of 1 variable:
.. ..$ ID: chr [1:31] "ARSON" "ASSAULT" "BATTERY" "BURGLARY" ...
..$ location:'data.frame': 135 obs. of 1 variable:
.. ..$ ID: chr [1:135] "" "ABANDONED BUILDING" "AIRCRAFT" "AIRPORT BUILDING NON-TERMINAL - NON-SECURE AREA" ...
..$ arrest :'data.frame': 2 obs. of 1 variable:
.. ..$ ID: chr [1:2] "N" "Y"
..$ domestic:'data.frame': 2 obs. of 1 variable:
.. ..$ ID: chr [1:2] "N" "Y"
..$ ward :'data.frame': 50 obs. of 1 variable:
.. ..$ ID: chr [1:50] "1" "2" "3" "4" ...
..$ fbi :'data.frame': 26 obs. of 1 variable:
.. ..$ ID: chr [1:26] "01A" "01B" "02" "03" ...
$ links :'data.frame': 239616 obs. of 7 variables:
..$ one : num [1:239616] 1 1 1 1 1 1 1 1 1 1 ...
..$ primary : int [1:239616] 30 9 30 30 30 30 30 30 30 6 ...
..$ location: int [1:239616] 120 103 27 105 33 19 120 62 120 64 ...
..$ arrest : int [1:239616] 1 1 1 1 1 1 1 1 1 1 ...
..$ domestic: int [1:239616] 1 1 1 1 1 1 1 1 1 1 ...
..$ ward : int [1:239616] 27 39 42 49 42 27 43 25 50 5 ...
..$ fbi : int [1:239616] 8 14 8 8 8 8 8 8 8 17 ...
$ data : list()
Some variables may need a special treatment - as date/time that we would like transform into hours.
> # https://en.wikipedia.org/wiki/12-hour_clock
> hour <- function(s){t <- unlist(strsplit(s," "))
+ as.integer(unlist(strsplit(t[2],":"))[1])%%12+12*(t[3]=="PM")}
> sapply(CH$date[1:20],hour)
01/29/2022 11:55:00 PM 01/29/2022 12:02:00 PM 01/29/2022 07:35:00 PM 03/10/2022 06:20:00 PM
23 12 19 18
03/12/2022 01:45:00 PM 01/29/2022 07:50:00 PM 01/25/2022 09:00:00 AM 08/07/2022 08:05:00 PM
13 19 9 20
02/21/2022 03:00:00 PM 03/12/2022 11:00:00 PM 09/06/2022 12:00:00 PM 01/25/2022 12:36:00 AM
15 23 12 0
03/24/2022 04:27:00 AM 03/13/2022 04:45:00 PM 11/03/2022 08:30:00 AM 02/22/2022 06:00:00 PM
4 16 8 18
03/23/2022 02:35:00 PM 03/15/2022 07:30:00 PM 05/13/2022 09:30:00 PM 03/13/2022 02:30:00 PM
14 19 21 14
We add such weights separately. We can add also some additional data or metadata.
> h <- sapply(CH$date,hour) > MT$links$hour <- h > MT$info$URL <- "https://www.chicago.gov/city/en/dataset/crime.html" > names(MT$links)[1] <- "w" > write(toJSON(MT),"ChicagoCrime0.json")
The basic multiway network constructed in the previous step can still have multiple links. To get a simple network we have to aggregate it. In this process, we can also omit some ways.
> L <- MT$links
> Links <- aggregate(L$w,by=list(L$primary,L$location,L$ward),FUN=sum)
> names(Links) <- c("primary","location","ward","w")
> MT$links <- Links
> ways <- c("primary","location","ward"); Ways <- as.list(ways); names(Ways) <- ways
> MT$ways <- Ways
> MT$nodes <- list(MT$nodes$primary,MT$nodes$location,MT$nodes$ward)
> names(MT$nodes) <- MT$ways
> MT$nodes$ward$long <- paste("ward",MT$nodes$ward$ID)
> str(MT)
List of 6
$ format: chr "MWnets"
$ info :List of 5
..$ network: chr "ChicagoCrime22"
..$ title : chr "City of Chicago incidents of crime 2022"
..$ by : chr "DF2MWN"
..$ date : chr "Mon Jan 30 04:02:07 2023"
..$ URL : chr "https://www.chicago.gov/city/en/dataset/crime.html"
$ ways :List of 3
..$ primary : chr "primary"
..$ location: chr "location"
..$ ward : chr "ward"
$ nodes :List of 3
..$ primary :'data.frame': 31 obs. of 1 variable:
.. ..$ ID: chr [1:31] "ARSON" "ASSAULT" "BATTERY" "BURGLARY" ...
..$ location:'data.frame': 135 obs. of 1 variable:
.. ..$ ID: chr [1:135] "" "ABANDONED BUILDING" "AIRCRAFT" "AIRPORT BUILDING NON-TERMINAL - NON-SECURE AREA" ...
..$ ward :'data.frame': 50 obs. of 2 variables:
.. ..$ ID : chr [1:50] "1" "2" "3" "4" ...
.. ..$ long: chr [1:50] "ward 1" "ward 2" "ward 3" "ward 4" ...
$ links :'data.frame': 18598 obs. of 4 variables:
..$ primary : int [1:18598] 9 6 27 30 1 2 3 6 8 9 ...
..$ location: int [1:18598] 1 2 2 2 17 17 17 17 17 17 ...
..$ ward : int [1:18598] 1 1 1 1 1 1 1 1 1 1 ...
..$ w : num [1:18598] 20 1 1 1 1 12 8 22 2 2 ...
$ data : list()
> write(toJSON(MT),"ChicagoCrime1.json")
January 30, 2023