projects
/
noahs_yelp_analysis
/ commitdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
| commitdiff |
tree
raw
|
patch
|
inline
| side by side (from parent 1:
ae54477
)
fixed errors (x-png > x.png | missing 2nd png(~~~) | non-hardcoded input)
master
author
Noah Swartz
<swartzcr@gmail.com>
Mon, 9 Apr 2012 20:02:49 +0000
(16:02 -0400)
committer
Benjamin Mako Hill
<mako@atdot.cc>
Mon, 9 Apr 2012 22:47:35 +0000
(18:47 -0400)
analysis.R
patch
|
blob
|
history
diff --git
a/analysis.R
b/analysis.R
index aadc02e6ccde4d9da620f181905b70c632b5f137..14efe8977d184f0b4c34d15d1a0f9607bee1d100 100644
(file)
--- a/
analysis.R
+++ b/
analysis.R
@@
-1,7
+1,7
@@
-
q
cat("\n","Enter textfile name","\n") # prompt
+cat("\n","Enter textfile name","\n") # prompt
y<-readLines(n=1)
y<-readLines(n=1)
-x <-
"lord-hobo-cambridge.txt"
+x <-
paste(y,".txt", sep="")
d <- read.csv(x)[,c(1,2,3)]
colnames(d) <- c("rating", "day.of.week", "day.in.review")
d <- read.csv(x)[,c(1,2,3)]
colnames(d) <- c("rating", "day.of.week", "day.in.review")
@@
-20,8
+20,8
@@
d$day.of.week <- factor(d$day.of.week, levels = c("Monday","Tuesday","Wednesday"
d$day.of.week <- as.factor(d$day.of.week)
d$day.of.week <- as.factor(d$day.of.week)
-t <- table(d$day.of.week, as.factor(d$rating))
-t2 <- prop.table(t,2)
+
#
t <- table(d$day.of.week, as.factor(d$rating))
+
#
t2 <- prop.table(t,2)
chisq.test(d$day.of.week, as.factor(d$rating))
# compute and add a set of weights so that we can graph proportions
chisq.test(d$day.of.week, as.factor(d$rating))
# compute and add a set of weights so that we can graph proportions
@@
-34,7
+34,7
@@
d <- merge(d, w, by="day.of.week", all.x=TRUE, all.y=FALSE)
library(ggplot2)
# generate the unscaled count graph
library(ggplot2)
# generate the unscaled count graph
-filename.count <- paste(y,"-count
-
png",sep="")
+filename.count <- paste(y,"-count
.
png",sep="")
png(filename.count, width=8, height=6, unit="in", res=200)
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar", main=y)
png(filename.count, width=8, height=6, unit="in", res=200)
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar", main=y)
@@
-45,10
+45,10
@@
print(p)
dev.off()
# generate the scaled proportion graph
dev.off()
# generate the scaled proportion graph
-filename.prop <- paste(y,"-proportion
-
png",sep="")
+filename.prop <- paste(y,"-proportion
.
png",sep="")
max.value <- tapply(d$plot.weight, d$day.of.week, sum)[1]
max.value <- tapply(d$plot.weight, d$day.of.week, sum)[1]
-
+png(filename.prop, width=8,height=6, unit="in", res=200)
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar",
weight=plot.weight, main=y)
p <- p + scale_x_discrete("Day of Week") +
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar",
weight=plot.weight, main=y)
p <- p + scale_x_discrete("Day of Week") +
Benjamin Mako Hill
||
Want to submit a patch?