india_blackout

india_blackout.R

library(ggplot2)
library(RPostgreSQL)
library(plyr)
library(scales) # now needed for ggplot2

drv <- dbDriver("PostgreSQL")
con <- dbConnect(drv, dbname='db',
                 host = 'localhost',
                 user = 'db',
                 password ='pwd',
                 port = '1234')

df.raw <- dbGetQuery(con, "select * from analytics.india_blackout_data") # get data from oDesk

df <- ddply(df.raw, .(date, obs_type), transform, # add standard errors
            se = sqrt(frac * (1-frac))/sqrt(n))

df.blackout <- subset(df, date == '2012-07-31') # to label blackout date
df.blackout$label  <- "Blackout"

df$weekend <- "" # annotate when the weekends are    
df$weekend[df$dow == 6] <- "Sat."
df$weekend[df$dow == 7] <- "Sun."

g <- ggplot(data = df, aes(x = date, y = frac)) +
  geom_errorbar(aes(ymin = frac - 2*se, ymax = frac + 2*se)) +
  geom_line(aes(x = date, y = frac, group = obs_type),linetype=3) +
  facet_wrap(~obs_type, ncol = 1) +
  geom_point(data = df.blackout, aes(x = date, y = frac), colour = "red", size = 4) +
  geom_text(aes(x = date, y = frac, label = weekend), 
                size = 4, hjust = 1.7, colour="blue") +
  geom_text(data = df.blackout, aes(x = date, y = frac, label = label), 
                colour = "red", vjust = 3, size = 4) + 
  scale_y_continuous(labels = percent) +
  ylab("Hours worked or applications sent by Indian contractors as a percentage of total") + 
  theme_bw()

print(g)
  
png("/home/john/Desktop/india_blackout.png", width = 600)
 print(g)
dev.off()