Using R programing [40 points] The dataset used in this problem is related to th
ID: 2267434 • Letter: U
Question
Using R programing
[40 points]
The dataset used in this problem is related to the quality of white wine. The data includes 4898 samples of wine. There are 11 variables have potential effect to the response variable (quality). The data file “winequality-white.txt” and its description “winequality-white_description.txt” are uploaded on Canvas. The goal of this problem is to get you familiar with R and using R-packages to make the decision tree model and random forest model.
Using “rpart” R-package to make a decision tree and plot the tree. Which variable has the highest impact to the quality of white wine (show your result)?
Making a random forest model using the whole dataset, report your results. How does this result look like in comparison to part a?
Data sources:
http://archive.ics.uci.edu/ml/datasets/Wine+Quality
https://www.superdatascience.com/
Explanation / Answer
# ---Simple Calculations---
2 + 3
x <- 2
y <- 3
x + y
x * y
# ---Data Structures---
# Vectors
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
print(workshop)
workshop
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
q1 <- c(1, 2, 2, 3, 4, 5, 5, 4)
q2 <- c(1, 1, 2, 1, 5, 4, 3, 5)
q3 <- c(5, 4, 4,NA, 2, 5, 4, 5)
q4 <- c(1, 1, 3, 3, 4, 5, 4, 5)
# Selecting Elements of Vectors
q1[5]
q1[ c(5, 6, 7, 8) ]
q1[5:8]
q1[gender == "m"]
mean( q1[ gender == "m" ], na.rm = TRUE)
# ---Factors---
# Numeric Factors
# First, as a vector
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop
table(workshop)
mean(workshop)
gender[workshop == 2]
# Now as a factor
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(workshop)
workshop
table(workshop)
mean(workshop) #generates error now.
gender[workshop == 2]
gender[workshop == "2"]
# Recreate workshop, making it a factor
# including levels that don't yet exist.
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(
workshop,
levels = c( 1, 2, 3, 4),
labels = c("R", "SAS", "SPSS", "Stata")
)
# Recreate it with just the levels it
# curently has.
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(
workshop,
levels = c( 1, 2),
labels = c("R","SAS")
)
workshop
table(workshop)
gender[workshop == 2]
gender[workshop == "2"]
gender[workshop == "SAS"]
# Character factors
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
gender <- factor(
gender,
levels = c("m", "f"),
labels = c("Male", "Female")
)
gender
table(gender)
workshop[gender == "m"]
workshop[gender == "Male"]
# Recreate gender and make it a factor,
# keeping simpler m and f as labels.
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
gender <- factor(gender)
gender
# Data Frames
mydata <- data.frame(workshop, gender, q1, q2, q3, q4)
mydata
names(mydata)
row.names(mydata)
# Selecting components by index number
mydata[8, 6] #8th obs, 6th var
mydata[ , 6] #All obs, 6th var
mydata[ , 6][5:8] #6th var, obs 5:8
# Selecting components by name
mydata$q1
mydata$q1[5:8]
# Example renaming gender to sex while
# creating a data frame (left as a comment)
#
# mydata <- data.frame(workshop, sex = gender,
# q1, q2, q3, q4)
# Matrices
# Creating from vectors
mymatrix <- cbind(q1, q2, q3, q4)
mymatrix
dim(mymatrix)
# Creating from matrix function
# left as a comment so we keep
# version with names q1, q2...
#
# mymatrix <- matrix(
# c(1, 1, 5, 1,
# 2, 1, 4, 1,
# 2, 2, 4, 3,
# 3, 1, NA,3,
# 4, 5, 2, 4,
# 5, 4, 5, 5,
# 5, 3, 4, 4,
# 4, 5, 5, 5),
# nrow = 8, ncol = 4, byrow = TRUE)
# mymatrix
table(mymatrix)
mean(mymatrix, na.rm = TRUE)
cor(mymatrix, use = "pairwise")
# Selecting Subsets of Matrices
mymatrix[8, 4]
mymatrix[5:8, 3:4]
mymatrix[ ,4][1:4]
mymatrix$q4 # No good!
mymatrix[ ,"q4"]
# Matrix Algebra
mymatrixT <- t(mymatrix)
mymatrixT
# Lists
mylist <- list(workshop, gender,
q1, q2, q3, q4, mymatrix)
mylist
# List, this time adding names
mylist <- list(
workshop = workshop,
gender = gender,
q1 = q1,
q2 = q2,
q3 = q3,
q4 = q4,
mymatrix = mymatrix)
mylist
# Selecting components by index number.
mylist[[2]]
mylist[[2]][5:8]
mylist[2]
mylist[2][5:8] # Bad!
# Selecting components by name.
mylist$q1
mylist$mymatrix[5:8, 3:4]
# ---Saving Your Work---
ls()
objects() #same as ls()
save.image("myall.RData")
save(mydata, file = "mydata.RData")
# The 2nd approach is commented to keep
# the q variables for following examples.
# rm(x, y, workshop, gender, q1, q2, q3, q4, mylist)
# ls()
# save.image(file = "mydata.RData")
# ---Comments to Document Your Programs---
# This comment is on its own line, between functions.
workshop <- c(1, 2, 1, 2, #This comment is within the arguments.
1, 2, 1, 2) # And this is at the end.
# ---Comments to Document Your Objects---
comment(mydata) <- "Example data from R for SAS and SPSS Users"
comment(mydata)
# ---Controlling Functions---
# Controlling Functions with Arguments
help("mean")
mean(x = q3, trim = .25, na.rm = TRUE)
mean(na.rm = TRUE, x = q3, trim = .25)
mean(q3, .25, TRUE)
mean(q3, t = .25, na.rm = TRUE)
mean(1, 2, 3)
mean( c(1, 2, 3) )
mean( 1:3 )
# Controlling Functions With Formulas
lm( q4 ~ q1 + q2 + q3, data = mydata )
t.test(q1 ~ gender, data = mydata)
t.test( q1[ which(gender == "Female") ],
q1[ which(gender == "Male") ],
data = mydata) # Data ignored!
# Controlling Functions with Extractor Functions
lm( q4 ~ q1 + q2 + q3, data = mydata )
myModel <- lm( q4 ~ q1 + q2 + q3, data = mydata )
class(myModel)
summary(myModel)
# How Much Output Is There?
print(mymodel)
mode(myModel)
class(myModel)
names(myModel)
print( unclass(myModel) )
myModel$coefficients
class( myModel$coefficients )
barplot( myModel$coefficients )
# ---Writing Your Own Functions (Macros)---
myvar <- c(1, 2, 3, 4, 5)
# A bad function.
mystats <- function(x) {
mean(x, na.rm = TRUE)
sd(x, na.rm = TRUE)
}
mystats(myvar)
# A good function that just prints.
mystats <- function(x) {
print( mean(x, na.rm = TRUE) )
print( sd(x, na.rm = TRUE) )
}
mystats(myvar)
# A function with vector output.
mystats <- function(x) {
mymean <- mean(x, na.rm = TRUE)
mysd <- sd(x, na.rm = TRUE)
c(mean = mymean, sd = mysd )
}
mystats(myvar)
myVector <- mystats(myvar)
myVector
# A function with list output.
mystats <- function(x) {
myinput <- x
mymean <- mean(x, na.rm = TRUE)
mysd <- sd(x, na.rm = TRUE)
list(data = myinput, mean = mymean, sd = mysd)
}
mystats(myvar)
myStatlist <- mystats(myvar)
myStatlist
mystats
save(mydata, mymatrix, mylist, mystats,
file = "myWorkspace.RData")
# ---Simple Calculations---
2 + 3
x <- 2
y <- 3
x + y
x * y
# ---Data Structures---
# Vectors
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
print(workshop)
workshop
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
q1 <- c(1, 2, 2, 3, 4, 5, 5, 4)
q2 <- c(1, 1, 2, 1, 5, 4, 3, 5)
q3 <- c(5, 4, 4,NA, 2, 5, 4, 5)
q4 <- c(1, 1, 3, 3, 4, 5, 4, 5)
# Selecting Elements of Vectors
q1[5]
q1[ c(5, 6, 7, 8) ]
q1[5:8]
q1[gender == "m"]
mean( q1[ gender == "m" ], na.rm = TRUE)
# ---Factors---
# Numeric Factors
# First, as a vector
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop
table(workshop)
mean(workshop)
gender[workshop == 2]
# Now as a factor
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(workshop)
workshop
table(workshop)
mean(workshop) #generates error now.
gender[workshop == 2]
gender[workshop == "2"]
# Recreate workshop, making it a factor
# including levels that don't yet exist.
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(
workshop,
levels = c( 1, 2, 3, 4),
labels = c("R", "SAS", "SPSS", "Stata")
)
# Recreate it with just the levels it
# curently has.
workshop <- c(1, 2, 1, 2, 1, 2, 1, 2)
workshop <- factor(
workshop,
levels = c( 1, 2),
labels = c("R","SAS")
)
workshop
table(workshop)
gender[workshop == 2]
gender[workshop == "2"]
gender[workshop == "SAS"]
# Character factors
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
gender <- factor(
gender,
levels = c("m", "f"),
labels = c("Male", "Female")
)
gender
table(gender)
workshop[gender == "m"]
workshop[gender == "Male"]
# Recreate gender and make it a factor,
# keeping simpler m and f as labels.
gender <- c("f", "f", "f", NA, "m", "m", "m", "m")
gender <- factor(gender)
gender
# Data Frames
mydata <- data.frame(workshop, gender, q1, q2, q3, q4)
mydata
names(mydata)
row.names(mydata)
# Selecting components by index number
mydata[8, 6] #8th obs, 6th var
mydata[ , 6] #All obs, 6th var
mydata[ , 6][5:8] #6th var, obs 5:8
# Selecting components by name
mydata$q1
mydata$q1[5:8]
# Example renaming gender to sex while
# creating a data frame (left as a comment)
#
# mydata <- data.frame(workshop, sex = gender,
# q1, q2, q3, q4)
# Matrices
# Creating from vectors
mymatrix <- cbind(q1, q2, q3, q4)
mymatrix
dim(mymatrix)
# Creating from matrix function
# left as a comment so we keep
# version with names q1, q2...
#
# mymatrix <- matrix(
# c(1, 1, 5, 1,
# 2, 1, 4, 1,
# 2, 2, 4, 3,
# 3, 1, NA,3,
# 4, 5, 2, 4,
# 5, 4, 5, 5,
# 5, 3, 4, 4,
# 4, 5, 5, 5),
# nrow = 8, ncol = 4, byrow = TRUE)
# mymatrix
table(mymatrix)
mean(mymatrix, na.rm = TRUE)
cor(mymatrix, use = "pairwise")
# Selecting Subsets of Matrices
mymatrix[8, 4]
mymatrix[5:8, 3:4]
mymatrix[ ,4][1:4]
mymatrix$q4 # No good!
mymatrix[ ,"q4"]
# Matrix Algebra
mymatrixT <- t(mymatrix)
mymatrixT
# Lists
mylist <- list(workshop, gender,
q1, q2, q3, q4, mymatrix)
mylist
# List, this time adding names
mylist <- list(
workshop = workshop,
gender = gender,
q1 = q1,
q2 = q2,
q3 = q3,
q4 = q4,
mymatrix = mymatrix)
mylist
# Selecting components by index number.
mylist[[2]]
mylist[[2]][5:8]
mylist[2]
mylist[2][5:8] # Bad!
# Selecting components by name.
mylist$q1
mylist$mymatrix[5:8, 3:4]
# ---Saving Your Work---
ls()
objects() #same as ls()
save.image("myall.RData")
save(mydata, file = "mydata.RData")
# The 2nd approach is commented to keep
# the q variables for following examples.
# rm(x, y, workshop, gender, q1, q2, q3, q4, mylist)
# ls()
# save.image(file = "mydata.RData")
# ---Comments to Document Your Programs---
# This comment is on its own line, between functions.
workshop <- c(1, 2, 1, 2, #This comment is within the arguments.
1, 2, 1, 2) # And this is at the end.
# ---Comments to Document Your Objects---
comment(mydata) <- "Example data from R for SAS and SPSS Users"
comment(mydata)
# ---Controlling Functions---
# Controlling Functions with Arguments
help("mean")
mean(x = q3, trim = .25, na.rm = TRUE)
mean(na.rm = TRUE, x = q3, trim = .25)
mean(q3, .25, TRUE)
mean(q3, t = .25, na.rm = TRUE)
mean(1, 2, 3)
mean( c(1, 2, 3) )
mean( 1:3 )
# Controlling Functions With Formulas
lm( q4 ~ q1 + q2 + q3, data = mydata )
t.test(q1 ~ gender, data = mydata)
t.test( q1[ which(gender == "Female") ],
q1[ which(gender == "Male") ],
data = mydata) # Data ignored!
# Controlling Functions with Extractor Functions
lm( q4 ~ q1 + q2 + q3, data = mydata )
myModel <- lm( q4 ~ q1 + q2 + q3, data = mydata )
class(myModel)
summary(myModel)
# How Much Output Is There?
print(mymodel)
mode(myModel)
class(myModel)
names(myModel)
print( unclass(myModel) )
myModel$coefficients
class( myModel$coefficients )
barplot( myModel$coefficients )
# ---Writing Your Own Functions (Macros)---
myvar <- c(1, 2, 3, 4, 5)
# A bad function.
mystats <- function(x) {
mean(x, na.rm = TRUE)
sd(x, na.rm = TRUE)
}
mystats(myvar)
# A good function that just prints.
mystats <- function(x) {
print( mean(x, na.rm = TRUE) )
print( sd(x, na.rm = TRUE) )
}
mystats(myvar)
# A function with vector output.
mystats <- function(x) {
mymean <- mean(x, na.rm = TRUE)
mysd <- sd(x, na.rm = TRUE)
c(mean = mymean, sd = mysd )
}
mystats(myvar)
myVector <- mystats(myvar)
myVector
# A function with list output.
mystats <- function(x) {
myinput <- x
mymean <- mean(x, na.rm = TRUE)
mysd <- sd(x, na.rm = TRUE)
list(data = myinput, mean = mymean, sd = mysd)
}
mystats(myvar)
myStatlist <- mystats(myvar)
myStatlist
mystats
save(mydata, mymatrix, mylist, mystats,
file = "myWorkspace.RData")
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.