---
title: "Flights Related Questions"
---

```{r}
#| label: setup
#| include: false
knitr::opts_chunk$set(
  echo = TRUE,
  message = FALSE,
  warning = FALSE,
  fig.width = 8,
  fig.height = 5,
  fig.align = "center",
  dpi = 120,
  error = TRUE,
  fig.path = "_figs/flights-practice/"
)
```

```{r}
library(tidyverse)
library(nycflights13)
```

### Question 1 — Flight Volume and Arrival Delays

Using the `flights` dataset, investigate whether the **number of flights an airline operates on a given day is related to its average arrival delay that day**. Restrict your analysis to airline-days with at least **20 flights** and exclude flights with missing arrival-delay information.

Create an appropriate visualization showing this relationship for the different airlines.

```{r}
#| fig-height: 6
flights %>%
  
  # Remove flights with missing arrival delay values
  filter(!is.na(arr_delay)) %>%
  
  # Group flights by airline and calendar day
  group_by(carrier, month, day) %>%
  
  # Calculate the number of flights and average arrival delay
  # for each airline on each day
  summarise(
    n_flights = n(),
    ave_arr_delay = mean(arr_delay),
    .groups = "drop"
  ) %>%
  
  # Keep only airline-days with at least 20 flights
  filter(n_flights >= 20) %>%
  
  # Sort from highest to lowest average arrival delay
  arrange(desc(ave_arr_delay)) %>%
  
  # Plot number of flights against average arrival delay
  ggplot(aes(
    x = n_flights,
    y = ave_arr_delay,
    colour = carrier
  )) +
  
  # Add scatter plot points with some transparency
  geom_point(alpha = 0.5) +
  
  # Add axis labels and a title
  xlab("Number of Flights") +
  ylab("Average Arrival Delay") +
  ggtitle("Number of Flights vs. Average Arrival Delay by Carrier") +
  
  # Create a separate panel for each airline
  facet_wrap(~ carrier)

```

### Question 2 — Worst Month for Each Airline

Using the `flights` dataset, determine the **worst month for each airline based on departure delays**. Define an airline’s worst month as the month having its **highest average departure delay**. Exclude flights with missing departure-delay information.

Create a bar chart comparing the airlines, where the **height of each bar represents the maximum monthly average departure delay** and the color identifies the corresponding worst month.

```{r}
flights %>%
  
  # Remove flights with missing departure delay values
  filter(!is.na(dep_delay)) %>%
  
  # Group flights by airline and month
  group_by(carrier, month) %>%
  
  # Calculate the average departure delay for each
  # airline in each month
  summarise(
    avg_dep_delay = mean(dep_delay),
    .groups = "drop"
  ) %>%
  
  # Group the monthly averages by airline
  group_by(carrier) %>%
  
  # Find the maximum monthly average departure delay
  # for each airline
  mutate(
    max_avg_dep_delay = max(avg_dep_delay)
  ) %>%
  
  # Keep only the month with the maximum average
  # departure delay for each airline
  filter(avg_dep_delay == max_avg_dep_delay) %>%
  
  # Remove the grouping
  ungroup() %>%
  
  # Create a bar plot of the worst monthly delay
  # for each airline
  ggplot(aes(
    x = carrier,
    y = max_avg_dep_delay,
    fill = factor(month)
  )) +
  
  # Create one bar for each airline
  geom_col() +
  
  # Add axis labels, legend title, and plot title
  labs(
    x = "Carrier",
    y = "Maximum Monthly Average Departure Delay (minutes)",
    fill = "Worst Month",
    title = "Worst Month for Each Airline Based on Departure Delay"
  ) +
  
  # Apply a minimal theme
  theme_minimal()
```

### Question 3 — What Is the Worst Working Day to Fly with Each Airline?

Using the `flights` dataset, determine the **worst working day of the week to fly with each airline based on departure delays**. Consider only **Monday through Friday**, and exclude flights for which departure-delay information is missing.

Define the **worst working day** for an airline as the weekday with the **highest average departure delay** for that airline.

Create a bar chart comparing the airlines, where the **height of each bar represents the maximum average departure delay across the five working days**, and the **color of the bar identifies the corresponding worst weekday**.

```{r}

flights %>%
  filter(!is.na(dep_delay)) %>%
  
  # Create date and weekday
  mutate(
    date = make_date(year, month, day),
    weekday = wday(date, label = TRUE, abbr = FALSE)
  ) %>%
  
  # Keep Monday through Friday
  filter(weekday %in% c(
    "Monday", "Tuesday", "Wednesday", "Thursday", "Friday"
  )) %>%
  
  # Average delay for each airline and weekday
  group_by(carrier, weekday) %>%
  summarise(
    avg_dep_delay = mean(dep_delay),
    .groups = "drop"
  ) %>%
  
  # Maximum weekday average for each airline
  group_by(carrier) %>%
  mutate(
    max_avg_dep_delay = max(avg_dep_delay)
  ) %>%
  
  # Keep the worst weekday
  filter(avg_dep_delay == max_avg_dep_delay) %>%
  ungroup() %>%
  
  # Plot
  ggplot(aes(
    x = carrier,
    y = max_avg_dep_delay,
    fill = weekday
  )) +
  geom_col() +
  labs(
    x = "Carrier",
    y = "Maximum Average Departure Delay (minutes)",
    fill = "Worst Weekday",
    title = "Worst Working Day to Fly for Each Airline"
  ) +
  theme_minimal()
```