week9-data-wrangling.R

## ----setup, include=FALSE---------------------------------------------------------------
knitr::opts_chunk$set(echo = TRUE)


## ----message=FALSE, warning=FALSE-------------------------------------------------------
library(dplyr)
library(tidyr)
library(ggplot2)


## ---------------------------------------------------------------------------------------
1:10 %>% sqrt() %>% mean() %>% round(digits=1) 


## ---------------------------------------------------------------------------------------
round(mean(sqrt(1:10)),digits=1) 


## ---------------------------------------------------------------------------------------
fun <- function(a=1, b=10){c(max(a), min(b))}
a <- 2:10
a %>% fun(a^2)


## ---------------------------------------------------------------------------------------
a <- 3
a %>% rnorm(n=a)
a %>% rnorm()


## ----message=FALSE----------------------------------------------------------------------
library(microbenchmark)
bm <- microbenchmark(1:10 %>% sqrt() %>% mean() %>% round(digits=1), round(mean(sqrt(1:10)),digits=1) )
ggplot2::autoplot(bm)


## ---------------------------------------------------------------------------------------
fun <- function(x, cyl=1){
x %>% transform(new = mpg^cyl) %>% select(mpg, new)
}


## ---------------------------------------------------------------------------------------
df <- data.frame(mpg=runif(10, 10, 30))
fun(df, cyl=0.5)


## ---------------------------------------------------------------------------------------
fun(mtcars[1:5,], cyl=0.5)


## ---------------------------------------------------------------------------------------
head(mtcars)


## ---------------------------------------------------------------------------------------
mtcars %>% select(mpg)


## ---------------------------------------------------------------------------------------
mtcars %>% select(-mpg, -qsec, -gear, -cyl)

## ---------------------------------------------------------------------------------------
mtcars %>% select(disp:qsec)


## ---------------------------------------------------------------------------------------
mtcars %>% select(!disp:qsec)

## ---------------------------------------------------------------------------------------
df %>% select(where(is.numeric))


## ---------------------------------------------------------------------------------------
mtcars %>% filter(cyl==4)


## ---------------------------------------------------------------------------------------
mtcars %>% filter(cyl==4, hp>90)


## ---------------------------------------------------------------------------------------
mtcars %>% 
  summarize(mean.wt=mean(wt))


## ---------------------------------------------------------------------------------------
mtcars %>% 
  summarize(mean.wt=mean(wt)) %>%
  round(digits=2)


## ---------------------------------------------------------------------------------------
round(mean(mtcars$wt), digits=2)


## ---------------------------------------------------------------------------------------
mtcars %>% 
  summarize(mean.wt=mean(c(mpg, hp)))


## ---------------------------------------------------------------------------------------
mean(c(mtcars$mpg, mtcars$hp))


## ---------------------------------------------------------------------------------------
mtcars %>% 
  summarize(mean.wt=mean(mpg:hp))


## ---------------------------------------------------------------------------------------
mtcars[1,"mpg"]:mtcars[1,"hp"]


## ---------------------------------------------------------------------------------------
mean(mtcars[1,"mpg"]:mtcars[1,"hp"])


## ---------------------------------------------------------------------------------------
mtcars %>% summarize_at(vars(mpg:hp), mean)


## ---------------------------------------------------------------------------------------
mtcars %>% 
  select(mpg:hp) %>% 
  summarize_all(mean)

## ---------------------------------------------------------------------------------------
apply(mtcars[,1:4],2,mean)


## ---------------------------------------------------------------------------------------
df <- mtcars[,1:3]; df$label="test"
df %>% summarize_all(mean)


## ---------------------------------------------------------------------------------------
apply(df,2,mean)


## ---------------------------------------------------------------------------------------
mtcars %>% 
  filter(cyl==4) %>% 
  select(mpg:hp) %>% 
  summarize_all(mean) %>%
  round(digits=2) %>%
  paste(collapse=" -- ")


## ---------------------------------------------------------------------------------------
df2 <- data.frame(id=paste0(rep("s",8),1:2),
                  pop=paste0(rep("r",8),rep(1:2,each=4)),
                  Week1=rnorm(8), Week2=rnorm(8), Week3=rnorm(8), Week4=rnorm(8))
head(df2)


## ---------------------------------------------------------------------------------------
df2 %>% group_by(pop) %>% summarize_all(mean)


## ---------------------------------------------------------------------------------------
apply(df2[df2$pop=="r1",3:6],2,mean)


## ---------------------------------------------------------------------------------------
df2 %>% 
  group_by(pop, id) %>%
  summarize(mean=mean(c(Week1, Week2, Week3, Week4)))


## ---------------------------------------------------------------------------------------
df2 %>% group_by(pop) %>% 
  select(where(is.numeric)) %>% 
  summarize_all(mean)


## ---------------------------------------------------------------------------------------
df2 %>% group_by(pop) %>% 
  select(ends_with("2")) %>% 
  summarize_all(mean)


## ---------------------------------------------------------------------------------------
mtcars %>% 
   select(mpg:wt) %>%
   mutate(wt.hp=wt/hp)


## ---------------------------------------------------------------------------------------
fun <- function(x,y){fit <- lm(y~x); coef(fit)[1]} #return interept
mtcars %>% 
  group_by(carb) %>% 
  summarize(lmm=fun(mpg,hp))


## ---------------------------------------------------------------------------------------
mtcars %>% 
  group_by(carb) %>% 
  mutate(lmm=fun(mpg,hp)) %>%
  select(mpg, hp, lmm)


## ---------------------------------------------------------------------------------------
mtcars %>% 
   select(mpg:wt) %>%
   mutate(mean=sqrt(wt))


## ---------------------------------------------------------------------------------------
mtcars %>% 
   select(mpg:wt) %>%
   mutate(mean=mean(wt))


## ---------------------------------------------------------------------------------------
mtcars %>% 
   select(mpg:wt) %>%
   mutate(mean1 = (wt+hp)/2, mean2=mean(c(wt,hp)))


## ----message=FALSE----------------------------------------------------------------------
bm <- microbenchmark(
  apply(mtcars,2,sum), 
  summarize_all(mtcars, sum)
  )
ggplot2::autoplot(bm)


## ----message=FALSE----------------------------------------------------------------------
bm <- microbenchmark(
   mtcars[mtcars$cyl==4,], 
   filter(mtcars, cyl==4),
    subset(mtcars, cyl==4)
   )
ggplot2::autoplot(bm)


## ---------------------------------------------------------------------------------------
df <- subset(mtcars, select=c(mpg, wt, hp, carb))
df$model <- rownames(mtcars)
head(df)


## ---------------------------------------------------------------------------------------
fit <- lm(mpg ~ wt + hp, data=df)
summary(fit)


## ---------------------------------------------------------------------------------------
plot(df$mpg, col="red", ylim=c(0,200))
points(df$wt, col="green")
points(df$hp, col="blue")


## ---------------------------------------------------------------------------------------
plot(df$carb, df$mpg, col="red", ylim=c(0,200))
points(df$carb, df$wt, col="green")
points(df$carb, df$hp, col="blue")


## ---------------------------------------------------------------------------------------
df.long <- data.frame()
a <- data.frame(model=df$model, val=df$mpg, name="mpg", carb=df$carb)
df.long <- rbind(df.long, a)
a <- data.frame(model=df$model, val=df$wt, name="wt", carb=df$carb)
df.long <- rbind(df.long, a)
a <- data.frame(model=df$model, val=df$hp, name="hp", carb=df$carb)
df.long <- rbind(df.long, a)
df.long


## ---------------------------------------------------------------------------------------
ggplot(df.long, aes(x=model, y=val, col=name)) + geom_point()


## ---------------------------------------------------------------------------------------
ggplot(df.long, aes(x=model, y=val, col=name)) + geom_point() +
  facet_wrap(~name)


## ---------------------------------------------------------------------------------------
df.long2 <- 
  pivot_longer(
    df,
    cols = c("mpg", "wt", "hp"),
    names_to = "name",
    values_to = "val"
  )


## ---------------------------------------------------------------------------------------
df.long2


## ---------------------------------------------------------------------------------------
ggplot(df.long, aes(x=model, y=val, col=name)) + geom_point() +
  facet_wrap(~name)


## ---------------------------------------------------------------------------------------
df.long$name <- factor(df.long$name, levels=c("mpg", "wt", "hp"))
ggplot(df.long, aes(x=model, y=val, col=name)) + geom_point() +
  facet_wrap(~name) +
  theme(axis.text.x=element_blank())


## ---------------------------------------------------------------------------------------
pivot_wider( 
    df.long,
    names_from = "name", 
    values_from = "val"
  )


## ----results="hide"---------------------------------------------------------------------
df.long %>%
  pivot_wider( 
    names_from = "name", 
    values_from = "val"
  )