####### Statistics Programming with R, Session 2 #######

# Part 1: Statistical Testing and Prediction
data(mtcars)
View(mtcars)
?mtcars

install.packages("ggplot2")
library(ggplot2)

ggplot(mtcars, aes(x = wt, y = mpg)) + geom_point()

cor(mtcars$wt, mtcars$mpg)

cor.test(mtcars$wt, mtcars$mpg)

ct <- cor.test(mtcars$wt, mtcars$mpg)
ct

ct$p.value
str(ct)

fit <- lm(mpg ~ wt, data = mtcars)
summary(fit)

predict(fit)

newcar <- data.frame(wt = 4.5)
predict(fit, newdata = newcar)

ggplot(mtcars, aes(x = wt, y = mpg)) + geom_point() + geom_smooth(method = "lm")

mfit <- lm(mpg ~ wt + am, mtcars)
summary(mfit)

mfit <- lm(mpg ~ I(wt + am), mtcars)

ggplot(mtcars, aes(x = wt, y = mpg, color = am)) + geom_point()

ggplot(mtcars, aes(x = factor(am), y = mpg)) + geom_boxplot()

mfit <- lm(mpg ~ wt + disp, mtcars)
summary(mfit)
mfit <- lm(mpg ~ wt + qsec, mtcars)
summary(mfit)

ggplot(mtcars, aes(x = wt, y = mpg, color = qsec)) + geom_point()
ggplot(mtcars, aes(x = wt, y = mpg, color = qsec)) + geom_point(size = 5) + scale_color_continuous(low = "blue", high = "yellow")

mfit <- lm(mpg ~ I(wt^2) + disp, mtcars)

mtcars$weight_squared <- mtcars$wt^2

mfit <- lm(mpg ~ weight_squared + disp, mtcars)

?nls

# Part 2 : Exploratory Data Analysis

install.packages("dplyr")
library(dplyr)

browseVignettes()

load(url("http://dgrtwo.github.io/files/undata-213.RData"))
dim(x)
View(x)

x <- tbl_df(x)
class(x)

select(x, rcid, date, vote)
select(x, rcid, date, vote, unres, uniquename)
select(x, rcid:unres, uniquename)
select(x, -uniquename)

select(x, rcid:unres, country = uniquename)

newx <- select(x, rcid:unres, country = uniquename)

f(g(h(x)))
h(x) %>% g() %>% f()

newx <- x %>% select(rcid:unres, country = uniquename)

newx <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8)

newx$country
  
gsub('"', '', newx$country)

newx <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8) %>%
  mutate(country = gsub('"', '', country))

newx <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8) %>%
  mutate(country = gsub('"', '', country), unres = gsub('"', '', unres))
 
#newx <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8) %>%
#  mutate(country = gsub('"', '', country), unres = gsub('"', '', unres), vote = vote*2)



newx <- newx %>% mutate(vote = factor(votes[vote]))

install.packages("tidyr")
library(tidyr)


# either

votes <-  c("Yes", "Abstain", "No")

newx <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8) %>%
  mutate(country = gsub('"', '', country), unres = gsub('"', '', unres)) %>%
  mutate(vote = factor(votes[vote]))

newx$year = as.numeric(sapply(newx$date, function(x){strsplit(x, split = '-')[[1]][1]}))
newx$month = as.numeric(sapply(newx$date, function(x){strsplit(x, split = '-')[[1]][2]}))
newx$day = as.numeric(sapply(newx$date, function(x){strsplit(x, split = '-')[[1]][3]}))

# or

votes <-  c("Yes", "Abstain", "No")

newx2 <- x %>% select(rcid:vote, country = uniquename) %>% filter(vote < 8) %>%
  mutate(country = gsub('"', '', country), unres = gsub('"', '', unres)) %>%
  separate(date, c("year", "month", "day")) %>% mutate(year = as.numeric(year), month = as.numeric(month), day = as.numeric(day)) %>%
  mutate(vote = factor(votes[vote]))
newx <- newx2

# everyone

votesumm <- newx %>% group_by(year) %>% summarize(numvotes = n()) 

votesumm <- newx %>% group_by(year) %>% summarize(numvotes = n(), yes = mean(vote == "Yes")) 

ggplot(votesumm, aes(x = year, y = yes)) + geom_line()
ggplot(votesumm, aes(x = year, y = yes)) + geom_line() + geom_smooth()

votesumm <- newx %>% group_by(year, country) %>% summarize(numvotes = n(), yes = mean(vote == "Yes")) 

sort(unique(votesumm$country))

chosen_countries <- c("Finland", "United States of America", "U.S.S.R", "Australia", "Germany", "Germany, East", "Germany, West")

country_summary <- votesumm %>% filter(country %in% chosen_countries)

ggplot(country_summary, aes(x = year, y = yes, color = country)) + geom_line()

ggplot(country_summary, aes(x = year, y = yes, color = country, fill = country)) + geom_line() + geom_smooth()

ggplot(country_summary, aes(x = year, y = yes, color = country, fill = country)) + geom_line() + geom_smooth() + 
  scale_color_brewer(palette = "Set1") + scale_fill_brewer(palette = "Set1")