projects.mako.cc - noahs_yelp_analysis/blob - analysis.R

   1 cat("\n","Enter textfile name","\n") # prompt
   2 y<-readLines(n=1)
   3 x <- paste("lord-hobo-cambridge.txt",sep="")
   4 d <- read.csv(x)[,c(1,2,3)]
   5
   6 colnames(d) <- c("rating", "day.of.week", "day.in.review")
   7 d$day.in.review <- d$day.in.review == "True"
   8
   9 d$day.of.week <- as.character(d$day.of.week)
  10 d$day.of.week[d$day.of.week == "0"] <- "Sunday"
  11 d$day.of.week[d$day.of.week == "1"] <- "Monday"
  12 d$day.of.week[d$day.of.week == "2"] <- "Tuesday"
  13 d$day.of.week[d$day.of.week == "3"] <- "Wednesday"
  14 d$day.of.week[d$day.of.week == "4"] <- "Thursday"
  15 d$day.of.week[d$day.of.week == "5"] <- "Friday"
  16 d$day.of.week[d$day.of.week == "6"] <- "Saturday"
  17 d$day.of.week <- factor(d$day.of.week, levels = c("Monday","Tuesday","Wednesday","Thursday","Friday","Saturday","Sunday"))
  18
  19 d$day.of.week <- as.factor(d$day.of.week)
  20
  21 t <- table(d$day.of.week, as.factor(d$rating))
  22 t2 <- prop.table(t,2)
  23 chisq.test(d$day.of.week, as.factor(d$rating))
  24
  25 # draw a graph
  26 library(ggplot2)
  27 v <- paste(y,"-baaaarplot.png",sep="")
  28 png(v, width=8, height=6, unit="in", res=200)
  29 p <- qplot(day.of.week, data=d, fill=as.factor(rating), main=y)
  30 #p <- qplot(levels(d$day.of.week),as.factor(names(d$rating)), t, data=data.frame(t,levels(d$day.of.week)))
  31 p <- p + scale_x_discrete("Day of Week") + scale_y_continuous("yelp # star review") + scale_fill_discrete("")
  32 print(p)
  33 dev.off()
  34
  35
  36