drewconway

# File-Name:       Scotch_Pref.R                 
# Date:            2009-11-29                                
# Author:          Drew Conway                                       
# Purpose:         Display one-dimensional item response for scotch whiskey preference
# Data Used:       whiskey, package=flexmix
# Packages Used:   zelig,ggplot
# Output File:     scotch_pref.png
# Data Output:     
# Machine:         Drew Conway's MacBook 

drewconway

library(geoR)
library(geoRglm)

# Perfrom MCMC simulations
model<-list(cov.pars = c(1, 1), beta = 1, family = "poisson")
mcmc.test<-mcmc.control(S.scale = 0.45, thin = 1)
test.tune<-glsm.mcmc(haiti.geo, model = model, mcmc.input=mcmc.test)
haiti.mcmc<-prepare.likfit.glsm(test.tune)
prior<- prior.glm.control(phi.prior = "fixed", phi = .1)

drewconway

library(ggplot2)
library(XML)

### Meetup topics word cloud ###

# Get the raw meetup description into a dataframe
raw_desc<-levels(read.table('descriptions.txt',sep="\n")$V1)
clean_strings<-function(s){
    low<-tolower(s)
    clean<-gsub("[[:punct:]\n]","",low)

drewconway

### Meetup activity history ###

joins<-read.csv("New_York_R_Statistical_Programming_Meetup_Groups_Joins.csv")
rsvp<-read.csv("New_York_R_Statistical_Programming_Meetup_RSVPs.csv")
activity<-read.csv("New_York_R_Statistical_Programming_Meetup_Total_and_Active_Members.csv")

# Merge the data into a single frame
all<-merge(joins,rsvp,by="Date",all=TRUE)
all<-merge(all,activity,by="Date",all=TRUE)
all_dates<-as.vector(all$Date)

drewconway

# From the FrumForum.com Tea Party survey we see the following result for 
# the questions: 

# Question 1: In approximate percentage terms, how much is the U.S. (federal) 
# government currently taking out of the U.S. economy in taxation?
#
# Mean:     42.06%
# STD:      19.06%
# Actual:   31.5%
# 

drewconway

# Visualize data and save
png("Tax_percentage.png",height=1000,width=1000,res=100)
ggplot(frum_data,aes(tax1))+stat_density()+geom_vline(aes(xintercept=31.5,colour="Actual Value"))+opts(title="In approximate percentage terms, how much is the U.S. (federal)
    government currently taking out of the U.S. economy in taxation?",plot.title=theme_text(size = 12))+xlab("Simulated Tea Party Survey Response Distribution")+ylab("Density")
dev.off()    

png("Tax_family.png",height=1000,width=1000,res=100)
ggplot(frum_data,aes(tax2))+stat_density()+geom_vline(aes(xintercept=7.5,colour="Actual Value"))+opts(title="How much federal income tax do you think a typical family
    earning $50,000 pays (in 1,000 dollars)?",plot.title=theme_text(size = 12))+xlab("Simulated Tea Party Survey Response Distribution")+ylab("Density")
dev.off()

drewconway

library(survival)
library(ggplot2)
library(Design)


### Test data generation ###

# Chi-square test
count<-1000
cs1<-rchisq(count,1)

drewconway

### Generate our data ###

# Time to seeing a penis
penis<-round(rchisq(30,2))
p_cens<-rep(1,30)
p_type<-rep("penis",30)
p_bind<-cbind(penis,p_cens,p_type)

# Time to seeing a lonely dude
lonely<-round(rchisq(40,1))

drewconway

### Perform analysis and make pretty pictures ###

png("surv_plot.png",height=1000,width=1000,res=100)
par(mfrow=c(2,2))
survplot(survfit(Surv(time,censor)~type,data=subset(surv_data,type=="lonely")),what="survival",conf="bands",xlab="Minutes")
title("Survival function for seeing a lonely guy")
survplot(survfit(Surv(time,censor)~type,data=subset(surv_data,type=="penis")),what="survival",conf="bands",xlab="Minutes",main="Survival function for seeing a penis")
title("Survival function for seeing a penis")
survplot(survfit(Surv(time,censor)~type,data=subset(surv_data,type=="drunk")),what="survival",conf="bands",xlab="Minutes",main="Survival function for seeing two or more drunk people")
title("Survival function for seeing two or more drunk people")

drewconway

#### The following code produces a crash of ggplot2 ####

# Load data
hashtag<-"rstats"
infochimps<-read.csv(paste(hashtag,"_infochimps.csv",sep=""))

# Produce plot
png(paste(hashtag,"_infochimps_metric.png",sep=""),height=800,width=800,res=100)
ic.plot<-ggplot(infochimps,aes(x=log(followers_count/friends_count),y=trstrank))+geom_text(aes(label=screen_name,color=tweet.hash,size=replies_out/replies_in))
ic.plot<-ic.plot+xlab(expression(log[frac(Followers,Friends)]))+ylab("Infochimps.org trstrank")+opts(title=paste("Key Actor Analysis for",hashtag," with Infochimps.org Data",sep=""))