library(igraph)
setwd('~/Documents/Violence Cascades/Raw Data/')


person_verts = get.data.frame(person,'vertices')
vics = person_verts$vic
vic_verts = person_verts[vics,14:19]
first_vics = apply(vic_verts,1,min,na.rm=T)


load('arrests.RData')
arr = arrests[order(arrests$date2),]
arr = arr[match(unique(arr$ir2),arr$ir2),]
arr = arr[match(rownames(person_verts),arr$ir2),]
arr = arr[vics,]
offset = 16802
first_arrests = as.numeric(arr$date2) - offset

lag = first_vics - first_arrests
hist(lag,100,col='')
h = hist(lag,150,col='#1f78b4',border=NA,axes=T,
         xlab='Days between first arrest and first gunshot victimization',
         ylab='Frequency',main=NULL,cex.axis=0.6,cex.lab=0.6)
box(lwd=1.1)

dobs = as.Date(arr$birth_date,format='%m/%d/%Y')
ages = as.numeric(start_date + first_vics - dobs)
plot(table(round(ages/365)))








#### OLD VERSION
##### Load arrest data
arrests <- read.csv("2006to2014arrests2.csv", header=T, colClass=c("character"))
arrests$ir2 <- paste("ir", arrests$ir_no)
sub.arrests <- subset(arrests, select=c(as.character("rd_no"), as.character('ir2'), 
                                        as.character('arrest_date')))
colnames(sub.arrests) <- c("events", "individuals", "dates")
sub.arrests$dates = as.Date(sub.arrests$dates,format='%m/%d/%Y')
sub.arrests = sub.arrests[order(sub.arrests$dates),]
# individuals <- unique(sub.arrests$individuals)
# events <- unique(sub.arrests$events)

##### Look at nonfatal victims
shootings <- read.csv("shooting-data-withdate2.csv", header = T)
victims = shootings[shootings$INV_PARTY_TYPE_CD=="VIC",]
victims = victims[!is.na(victims$IR_NO),]
victims$ir2 <- paste("ir", victims$IR_NO)
victims$INCIDENT_DATE = as.Date(victims$INCIDENT_DATE,format='%m/%d/%y')
victims = victims[order(victims$INCIDENT_DATE),] # sort so match gets first infection
victims = victims[victims$ir2 %in% arrests$ir2,] # only look at vics who were arrested

vic_dates = victims$INCIDENT_DATE[match(unique(victims$ir2),victims$ir2)]
arrest_dates = sub.arrests$dates[match(unique(victims$ir2), sub.arrests$individuals)]
sum(vic_dates==arrest_dates)
# the number of vics in arrest table doesn't match number of nonfatal vics in person?

##### Look at fatal victims
murders = read.csv("murder-victims-13nov.csv", header=T)
murders = murders[!is.na(murders$VICTIM_IR_NO),]
murders = murders[murders$INJURY_DESCR=="SHOT",]
murders = murders[match(unique(murders$VICTIM_IR_NO),murders$VICTIM_IR_NO),]
murders$ir2 = paste("ir", murders$VICTIM_IR_NO)
start_date = as.Date("2005-12-31")
murders$INJURY_DATE = as.Date(murders$INJURY_DATE,format='%m/%d/%y')
murders = murders[murders$INJURY_DATE>=start_date,]
murders = murders[order(murders$INJURY_DATE),]
murders = murders[murders$ir2 %in% arrests$ir2,]

mur_dates = murders$INJURY_DATE
arrest_dates = sub.arrests$dates[match(murders$ir2, sub.arrests$individuals)]
sum(mur_dates==arrest_dates)