#!/usr/local/bin/R
# Create_Read_CSV.R
# Let us try to create and read CSV

# Load tidyverse
library(tidyverse)

# Get current working directory
currentdir <- getwd()
currentdir

# Create new working directory
newdir <- paste0(currentdir,"/HW_Answers", sep = "")
dir.create(newdir, recursive = TRUE)

# Set working directory
setwd(newdir)

# You can check current directory
getwd()

# Our goal is to create a CSV file with the following and calculate the sums:
#                   Placebo Vaccine
# Infected          162     8
# Not Infected      18163   18190

# Create a data frame
Placebo <- c(162, 18163)
Vaccine <- c(8, 18190)
vacctrial <- data.frame(Placebo, Vaccine)
vacctrial
rownames(vacctrial)
rownames(vacctrial) <- c("Infected", "Not_Infected")
rownames(vacctrial)
vacctrial
infsum <- sum(vacctrial[1,])
notinfsum <-sum(vacctrial[2,])
psum <- sum(vacctrial[,1])
vsum <- sum(vacctrial[,2])
allsum <- psum + vsum
vacctrial$Sum <- c(infsum, notinfsum)
vacctrial
vacctrial <- rbind(vacctrial, c(psum, vsum, allsum))
rownames(vacctrial) <- c("Infected", "Not_Infected", "Sum2")
vacctrial

# or you can put it in one command
vacctrial2 <- data.frame(Placebo = c(162, 18163), Vaccine =c(8, 18190))
rownames(vacctrial2) <- c("Infected", "Not_Infected")
vacctrial2
vacctrial2 <- as.table(vacctrial2)
infsum <- sum(vacctrial2[1,])
notinfsum <-sum(vacctrial2[2,])
psum <- sum(vacctrial2[,1])
vsum <- sum(vacctrial2[,2])
allsum <- psum + vsum
vacctrial2$Sum <- c(infsum, notinfsum)
vacctrial2
vacctrial2 <- rbind(vacctrial2, c(psum, vsum, allsum))
rownames(vacctrial2) <- c("Infected", "Not_Infected", "Sum2")
vacctrial2

# or you can create a matrix
vacctrial3 <- matrix(c(162, 18163, 8, 18190), nrow=2, ncol=2)
vacctrial3
rownames(vacctrial3) <- c("Infected", "Not_Infected")
colnames(vacctrial3) <- c("Placebo", "Vaccine")
vacctrial3
vacctrial3 <- as.table(vacctrial3)
vacctrial3 <- addmargins(vacctrial3)
vacctrial3

dir()    # There is nothing since there is no file in the newly created directory
# Let us create the CSV file (all three files below should be the same)
write.csv(vacctrial, file = paste0(newdir,"/Vaccine_Trial.csv"))
write.csv(vacctrial2, file = paste0(newdir,"/Vaccine_Trial2.csv"))
write.csv(vacctrial3, file = paste0(newdir,"/Vaccine_Trial3.csv"))

# Now let us read the CSV file
trialres <- read.csv(file = paste0(newdir,"/Vaccine_Trial.csv"))
trialres
trialres2 <- read.csv(file = paste0(newdir,"/Vaccine_Trial2.csv"))
trialres2
trialres3 <- read.csv(file = paste0(newdir,"/Vaccine_Trial3.csv"))
trialres3

# See the difference when you read it, I will use trialres here
# To avoid this, you can also use row.names = FALSE when you write the CSV file
# e.g., write.csv(vacctrial, file = paste0(newdir,"/Vaccine_Trial.csv"), row.names = FALSE)
trialres[,1]
trialres[,2]
trialres[,3]

infp <- as.numeric(trialres[1,2])
notinfp <- as.numeric(trialres[2,2])
totalp <- infp + notinfp
totalp

infv <- as.numeric(trialres[1,3])
notinfv <- as.numeric(trialres[2,3])
totalv <- infv + notinfv
totalv

riskv <- (infv/totalv) * 100
riskv
riskp <- (infp/totalp) * 100
riskp

# Relative Risk (%)
rel_risk <- (riskv/riskp) * 100
rel_risk

# Attributable Risk (%)
att_risk <- (riskp-riskv)
att_risk

# Number Needed to Treat
nnt <- 1/(att_risk/100)
round(nnt)