Module 8 – Data Manipulation with dplyr

Load and Prepare Data

setwd("~/Downloads/Intro to R/Module 8")
theoph <- read.csv("theoph.csv")
theoph$Subject <- factor(theoph$Subject)

Select Relevant Variables

df_subject_time_conc <- theoph |> 
  select(Subject, Time, conc)
df_subject_info <- theoph |> 
  select(Subject, Wt, Dose) |> 
  distinct()

Filter for Specific Subjects

first_subject <- theoph |> 
  filter(Subject == levels(theoph$Subject)[1])

first_four_subjects <- theoph |> 
  filter(Subject %in% levels(theoph$Subject)[1:4])

Summary Statistics

avg_conc_per_subject <- theoph |> 
  group_by(Subject) |> 
  summarise(avg_conc = mean(conc, na.rm = TRUE))
avg_conc_per_subject

## # A tibble: 12 × 2
##    Subject avg_conc
##    <fct>      <dbl>
##  1 1           6.44
##  2 2           4.82
##  3 3           5.09
##  4 4           4.94
##  5 5           5.78
##  6 6           3.53
##  7 7           3.91
##  8 8           4.27
##  9 9           4.89
## 10 10          5.93
## 11 11          4.51
## 12 12          5.41

Line Plot – First Subject

ggplot(first_subject, aes(x = Time, y = conc)) +
  geom_line() +
  ggtitle("Concentration Over Time – First Subject") +
  theme_minimal()

Line Plot – All Subjects Colored

ggplot(df_subject_time_conc, aes(x = Time, y = conc, color = Subject)) +
  geom_line() +
  ggtitle("Concentration Over Time by Subject") +
  theme_minimal()

Faceted Plot by Subject

ggplot(df_subject_time_conc, aes(x = Time, y = conc)) +
  geom_line() +
  facet_wrap(~ Subject) +
  ggtitle("Faceted Plot: Concentration Over Time by Subject") +
  theme_minimal()