+
+
+
+
+
+
+Data-Wrangling-Visualization
+Nicole Schlosberg
+September 29, 2020
+ +
+
+Part I
+
+
+Data Wrangling
+In the hackathon a project was proposed to collect data from student video watching, a sample of this data is available in the file video-data.csv.
+#library(tidyverse)
+library(tidyr)
+library(dplyr)##
+## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
+##
+## filter, lag## The following objects are masked from 'package:base':
+##
+## intersect, setdiff, setequal, unionD1 <- read.csv("video-data.csv", header = TRUE)
+
+#Create a data frame that only contains the years 2018
+D2 <- filter(D1, year == 2018)
+
+Histograms
+#Generate a histogram of the watch time for the year 2018
+hist(D2$watch.time)
#Change the number of breaks to 100, do you get the same impression?
+hist(D2$watch.time, breaks = 100)
#Cut the y-axis off at 10
+hist(D2$watch.time, breaks = 100, ylim = c(0,10))
#Restore the y-axis and change the breaks so that they are 0-5, 5-20, 20-25, 25-35
+hist(D2$watch.time, breaks = c(0,5,20,25,35))
+
+Plots
+#Plot the number of confusion points against the watch time
+plot(D1$confusion.points, D1$watch.time)
#Create two variables x & y
+x <- c(1,3,2,7,6,4,4)
+y <- c(2,4,2,3,2,4,3)
+
+#Create a table from x & y
+table1 <- table(x,y)
+
+#Display the table as a Barplot
+barplot(table1)
#Create a data frame of the average total key points for each year and plot the two against each other as a lines
+D3 <- D1 %>% group_by(year) %>% summarise(mean_key = mean(key.points))## `summarise()` ungrouping output (override with `.groups` argument)plot(D3$year, D3$mean_key, type = "l", lty = "dashed")
#Create a boxplot of total enrollment for three students
+D4 <- filter(D1, stid == 4|stid == 20| stid == 22)
+#The drop levels command will remove all the schools from the variable with no data
+D4 <- droplevels(D4)
+boxplot(D4$watch.time~D4$stid, xlab = "Student", ylab = "Watch Time")
+
+Pairs
+#Use matrix notation to select columns 2, 5, 6, and 7
+D5 <- D1[,c(2,5,6,7)]
+#Draw a matrix of plots for every combination of variables
+pairs(D5)
+
+Part II
+-
+
- Create a simulated data set containing 100 students, each with a score from 1-100 representing performance in an educational game. The scores should tend to cluster around 75. Also, each student should be given a classification that reflects one of four interest groups: sport, music, nature, literature. +
#rnorm(100, 75, 15) creates a random sample with a mean of 75 and standard deviation of 15
+#filter() can be used to set max min value and can only work with a data frame, for rows
+#select() for columns
+#round() rounds numbers to whole number values
+#sample() draws a random samples from the groups vector according to a uniform distribution
+
+score <- rnorm(100,75,15)
+hist(score,breaks = 30)
S1 <- data.frame(score)
+
+#Top and tail the scores
+S1 <- filter(S1, score <= 100)
+hist(S1$score)
S2 <- data.frame(rep(100,5)) #repeat 100 5 times and names the column a random name that is not helpful so use the names() command to rename
+names(S2) <- "score"
+S3 <- bind_rows(S1,S2) #must make sure that the names of the columns and the type match
+
+#S3$score <- ifelse(S3$score >= 100, 100, S3$score)
+
+S3$score <-round(S3$score,0)
+
+interest <- c("sport", "music", "nature", "liturature")
+S3$interest <- sample(interest, 100, replace = TRUE)
+S3$stid <- seq(1,100,1)-
+
- **Using base R commands, draw a histogram of the scores. Change the breaks in your histogram until you think they best represent your data. +
hist(S3$score, breaks = 10)
-
+
- Create a new variable that groups the scores according to the breaks in your histogram. +
#cut() divides the range of scores into intervals and codes the values in scores according to which interval they fall. We use a vector called `letters` as the labels, `letters` is a vector made up of the letters of the alphabet.
+label <- letters[1:10]
+S3$breaks <- cut(S3$score, breaks = 10, labels = label)-
+
- Now using the colorbrewer package (RColorBrewer; http://colorbrewer2.org/#type=sequential&scheme=BuGn&n=3) design a pallette and assign it to the groups in your data on the histogram. +
library(RColorBrewer)
+#Let's look at the available palettes in RColorBrewer
+display.brewer.all()
#The top section of palettes are sequential, the middle section are qualitative, and the lower section are diverging.
+#Make RColorBrewer palette available to R and assign to your bins
+
+S3$colors <- brewer.pal(10, "BrBG")
+#Use named palette in histogram
+hist(S3$score, col = S3$colors)
-
+
- Create a boxplot that visualizes the scores for each interest group and color each interest group a different color. +
#Make a vector of the colors from RColorBrewer
+interest.col <- brewer.pal(4, "BuPu")
+boxplot(score ~ interest, S3, col = interest.col)
-
+
- Now simulate a new variable that describes the number of logins that students made to the educational game. They should vary from 1-25. +
S3$login <- sample(1:25, 100, replace = TRUE)-
+
- Plot the relationships between logins and scores. Give the plot a title and color the dots according to interest group. +
plot(S3$score, S3$login, main = "Login vs. Score", xlab = "Score", ylab = "Login", col = interest.col) 
-
+
- R contains several inbuilt data sets, one of these in called AirPassengers. Plot a line graph of the the airline passengers over time using this data set. +
plot(AirPassengers, type = "l", xlab = "Date", ylab = "Passenger numbers") 
-
+
- Using another inbuilt data set, iris, plot the relationships between all of the variables in the data set. Which of these relationships is it appropriate to run a correlation on? +
plot(iris)
plot(iris$Sepal.Length,iris$Sepal.Width)
plot(iris$Petal.Length,iris$Petal.Width)
plot(iris$Petal.Length,iris$Sepal.Length)
plot(iris$Petal.Width,iris$Sepal.Width)
plot(iris$Petal.Width,iris$Sepal.Length)
plot(iris$Petal.Length,iris$Sepal.Width)
plot(iris$Species,iris$Sepal.Width, xlab = "Species", ylab = "Sepal Width")
plot(iris$Species,iris$Sepal.Length, xlab = "Species", ylab = "Sepal Length")
plot(iris$Species,iris$Petal.Width, xlab = "Species", ylab = "Petal Width")
plot(iris$Species,iris$Petal.Length, xlab = "Species", ylab = "Petal Length")
#Which of these relationships is it appropriate to run a correlation on?
+#Correlation between Sepal Length and Width
+corOfSepalLW <- cor(iris$Sepal.Length, iris$Sepal.Width)
+#Correlation between Petal Length and Width
+corOfPetalLW <- cor(iris$Petal.Length, iris$Petal.Width)
+#Correlation between Petal Length and Sepal Length
+corOfLengthPS <- cor(iris$Petal.Length, iris$Sepal.Length)
+#Correlation between Petal Width and Sepal Width
+corOfWidthPS <- cor(iris$Petal.Width, iris$Sepal.Width)
+
+
+
+
+
+Part III - Analyzing Swirl
+
+
+Data
+In this repository you will find data describing Swirl activity from the class so far this semester. Please connect RStudio to this repository.
+
+
+Instructions
+-
+
Insert a new code block
+Create a data frame from the
swirl-data.csvfile calledDF1
+
The variables are:
+course_name - the name of the R course the student attempted
+lesson_name - the lesson name
+question_number - the question number attempted correct - whether the question was answered correctly
+attempt - how many times the student attempted the question
+skipped - whether the student skipped the question
+datetime - the date and time the student attempted the question
+hash - anonymyzed student ID
-
+
Create a new data frame that only includes the variables
hash,lesson_nameandattemptcalledDF2
+Use the
group_byfunction to create a data frame that sums all the attempts for eachhashby eachlesson_namecalledDF3
+
#2
+DF1 <- read.csv("swirl-data.csv", header = TRUE)
+
+#3
+DF2<- select(DF1, hash, lesson_name, attempt)
+
+#4
+DF3 <- DF2 %>% group_by(hash, lesson_name) %>% summarise(sum_key = sum(attempt), .groups = "keep")-
+
On a scrap piece of paper draw what you think
DF3would look like if all the lesson names were column names
+Convert
DF3to this format
+
#6
+#Get rid of the NAs so the next step does not throw error and add extra column of NAs
+DF3 <- na.omit(DF3)
+DF3 <- spread(DF3, lesson_name, sum_key) -
+
Create a new data frame from
DF1calledDF4that only includes the variableshash,lesson_nameandcorrect
+Convert the
correctvariable so thatTRUEis coded as the number1andFALSEis coded as0
+
#7
+DF4 <- select(DF1, hash, lesson_name, correct)
+
+#8
+#Correct misspelled FALS at line 809 in swirl_data.csv and subsequent DF4 dataframe from DF1
+DF4$correct <- ifelse(DF4$correct == "FALS", "FALSE", DF4$correct)
+
+#Convert the chr that was created with last back to logi
+DF4$correct <- type.convert(DF4$correct)
+
+#Get rid of the NAs so the next steps do not throw "NAs introduced by coercion"
+DF4 <- DF4[complete.cases(DF4$correct),]
+
+#Converts logi to num so 0s and 1s instead of FALSE and TRUE
+DF4$correct <- as.numeric(DF4$correct) -
+
- Create a new data frame called
DF5that provides a mean score for each student on each course
+
#9
+DF5 <- DF4 %>% group_by(hash, lesson_name) %>% summarise(mean_key = mean(correct), .groups = "keep")-
+
- Convert the
datetimevariable into month-day-year format and create a new data frame (DF6) that shows the average correct for each day
+
#10
+DF6 <- select(DF1, hash, lesson_name, datetime, correct)
+
+#steps to get TRUE/FALSE to 1/0
+DF6$correct <- ifelse(DF6$correct == "FALS", "FALSE", DF6$correct)
+DF6$correct <- type.convert(DF6$correct)
+DF6 <- DF6[complete.cases(DF6$correct),]
+DF6$correct <- as.numeric(DF6$correct)
+
+#Creating average correct for each day
+DF6 <- DF6 %>% group_by(hash, datetime, correct) %>% summarise(meanByDay = mean(correct), .groups = "keep")
+
+#Convert 'datetime' to month-day-year by converting the parsed num***
+#library(lubridate)
+#dateConverted <- mdy_hms(DF6$datetime)
+#DF6 <- separate_rows(DF6, DF6$datetime, sep = "0")