Fred Truman
Design & Code Creative
Smoking, Drinking and Peeing in NYC: Oct. 1 – Oct. 14, 2011
October 24, 2011
Our assignment for this week was simply to explore some data. Wow. Pretty open-ended but enough of a kick to just get my hands dirty. I was going to use some NBA stats I had available in CSV, but because I have other classes I thought it might be a good idea to relate the data exploration to one of the topics I’m already invested in. So I went digging on NYC’s data site and decided to have a go with the year+ worth of 3-1-1 call data.
I downloaded the data in CSV format, and it was ~1.53GB. @_@ Wow. That’s a lot of data. So much so that I really couldn’t even open the file, let alone work with it. So I slowly went in manually and sliced out the first two weeks of October and decided that would be enough. In only two weeks there were 63,299 incident reports across 148 unique categories (e.g. “Noise”, or “Graffiti”, or even “Illegal Animal Kept as Pet”).
Before I got to plotting I was trying to figure out whether or not there exists granular enough resolution in the R maps library to get an NYC map and when I discovered that it didn’t I went searching for a library that might. To my frustration I couldn’t find anything, but then I realized if I just plot enough incidents according to latitude and longitude, the structure of the city would appear. I also had to parse out only the relevant information, which in my opinion the most interesting parameters in here are the locations of the incidents and the type of incident. So I created a list of only the interesting bits so that it would be easier to work with (“Borough” wasn’t really necessary here, btw):
threeoneone <- read.csv("311-succinct-02.csv", header=TRUE, stringsAsFactors=FALSE)
borough <- threeoneone["Borough"]
type <- threeoneone["Complaint.Type"]
lon <- threeoneone["Longitude"]
lat <- threeoneone["Latitude"]
incidents <- cbind(type,borough,lat,lon)
From this I could plot all 63,299 incidents, which I did (in black). Then I wanted to get in and find out more about the different types of complaints, so I made this guy:
unique.types <- (unique(type[[1]]))
and just took a look at the craziness in there and started pulling the categories out into their own lists and plotting them over the top:
youth.incidents <- incidents[incidents[[1]]=="Disorderly Youth",] bike.incidents <- incidents[incidents[[1]]=="Derelict Bicycle",] noise.incidents <- incidents[incidents[[1]]=="Noise",] urine.incidents <- incidents[incidents[[1]]=="Urinating in Public",] smoking.incidents <- incidents[incidents[[1]]=="Smoking",] drinking.incidents <- incidents[incidents[[1]]=="Drinking",]
I decided that I would check for a geographical correlation between drinking, smoking and peeing - although it doesn't look like there's anything binding them together. That's really as far as I've gotten, but I'm looking forward to playing around some more with such a rich set of data! (entire code is below)
threeoneone <- read.csv("311-succinct-02.csv", header=TRUE, stringsAsFactors=FALSE)
borough <- threeoneone["Borough"]
type <- threeoneone["Complaint.Type"]
lon <- threeoneone["Longitude"]
lat <- threeoneone["Latitude"]
unique.types <- (unique(type[[1]]))
incidents <- cbind(type,borough,lat,lon)
youth.incidents <- incidents[incidents[[1]]=="Disorderly Youth",]
bike.incidents <- incidents[incidents[[1]]=="Derelict Bicycle",]
noise.incidents <- incidents[incidents[[1]]=="Noise",]
urine.incidents <- incidents[incidents[[1]]=="Urinating in Public",]
smoking.incidents <- incidents[incidents[[1]]=="Smoking",]
drinking.incidents <- incidents[incidents[[1]]=="Drinking",]
plot_colors <- c("#FF000060","#d200ff60","#FFC60060","#FFFFFF10","#0000FF40")
legend_colors <- c("#000000","#FF0000","#d200ff","#FFC600","#FFFFFF","#0000FF")
par(bg = "gray40")
par(fg = "gray60")
par(col.lab = "gray60")
par(col.main = "gray60")
plot(incidents[[4]],incidents[[3]], pch=16, col="#00000002", bty="n", xlab="", xaxt = "n", yaxt = "n", ylab="", main="311 Calls NYC")
points(noise.incidents[[4]],noise.incidents[[3]], pch=16, col=plot_colors[4])
points(bike.incidents[[4]],bike.incidents[[3]], pch=16, col=plot_colors[1])
points(youth.incidents[[4]],youth.incidents[[3]], pch=16, col=plot_colors[2])
points(urine.incidents[[4]],urine.incidents[[3]], pch=16, col=plot_colors[3])
points(smoking.incidents[[4]],smoking.incidents[[3]], pch=16, col=plot_colors[5])
legend("topleft", c("All 311 calls","Derelict Bicycle","Disorderly Youth","Urinating in Public","Noise", "Smoking"), cex=0.6, bty="n", pch=16, col=legend_colors);
#SMOKE DRINK PEE
smoke_drink_pee_colors <- c("#d200ff60","#00ff4860","#fcff0060")
smoke_drink_pee_legend <- c("#000000","#d200ff","#00ff48","#fcff00")
plot(incidents[[4]],incidents[[3]], pch=16, col="#00000002", bty="n", xlab="", xaxt = "n", yaxt = "n", ylab="", main="311 Calls NYC")
points(smoking.incidents[[4]],smoking.incidents[[3]], pch=16, col=smoke_drink_pee_colors[1])
points(drinking.incidents[[4]],drinking.incidents[[3]], pch=16, col=smoke_drink_pee_colors[2])
points(urine.incidents[[4]],urine.incidents[[3]], pch=16, col=smoke_drink_pee_colors[3])
legend("topleft", c("All 311 calls","Smoking","Drinking","Urinating in Public"), cex=0.8, bty="n", pch=16, col=smoke_drink_pee_legend);