project3.R

gistfile1.txt

library(ggplot2)
source('helpers.R')

load('data.Rda')
head(data)

####### Setup data

data$crm <- (data$crimes / data$popul) * 1000
data$crm <- ceiling(data$crm)

data$region <- factor(data$region, 
                      labels=c("Northeast","Midwest","South","West"), 
                      levels=c(1,2,3,4))

####### 1

p1 <- ggplot(data, aes(crm, fill=region)) +  theme_classic() + 
  geom_histogram(binwidth = 20, position = "dodge") + scale_fill_brewer(palette="Set2") + 
  xlim(0, 200)
  
p


# plot crm as a function of region
data$n <- 1
agg<-aggregate(data[,c("crm", "n")],by=list(region=data$region),FUN=sum)
agg


p2 <- ggplot(agg, aes(region, crm, fill=region)) + 
  theme_classic() + 
  geom_col() +
  scale_fill_brewer(palette="Set2")
p

# calculate region
agg$mu <- agg$crm / agg$n
agg


#### 2

# lets plot the regular distribution against the poisson distribution


p <- list()
i = 1

for (region in levels(data$region)){
  
  # cut out the subsection which matches the region name
  df <- data[(data$region == region), ]
  
  # calculate the mean and response of the poisson for the different
  # crime levels in our data
  mu <- mean(df$crm)
  df$po <- dpois(df$crm, mu)
  
  # add the plot to the list of plots
  p[[i]] <- ggplot(df, aes(crm)) + theme_classic() + theme(plot.title = element_text(hjust = 0.5)) +
    geom_histogram(aes(y = ..density.., fill="Red"), binwidth = 5) +
    geom_ribbon(aes(ymin = 0, ymax = po, fill="Blue", alpha = 0.5)) +
    xlim(0,130) + ggtitle(region) + theme(legend.position = "none")
  
  i = i + 1
}

multiplot(p[[1]],p[[2]],p[[3]],p[[4]],cols=2)