DATA101-flightEDA

#----------------------------------------------
# DATA 101 - HOMEWORK 2
# 
#----------------------------------------------

# Please save this script as "YOUR LAST NAME_HW2.R" and upload the script to Canvas. 
# You should also upload a word document containing your write up and graphs.
# Please type your code into the sections outlined below. 

#----------------------------------------------
# Question 1

# This simplifies the code by shaving off the "fat" of the data we're looking to
# analyze. In other words, it cleans the data and enable us to more accurately 
# view the information that we want to work from.

#----------------------------------------------
# Question 2

# The data:

library(tidyverse)
install.packages("nycflights13")
library(nycflights13)
flights
View(flights)

#Calculating for the delay:

avg_delay <- summarise(by_day, delay = mean(dep_delay, na.rm = TRUE))

# Flights not cancelled:

not_cancelled <- flights %>%
  filter(!is.na(dep_delay), !is.na(arr_delay))

# If 328,063 weren't cancelled of 336,776, I can quick do that math:

327346/336776 # = 97.19992% not cancelled, which leaves ~ 2.59% or:
336776-327346
# 9430 cancelled. My proportion of cancelled flights should match that number.

cancelled_per_day <- 
  flights %>%
  mutate(cancelled = (is.na(arr_delay) | is.na(dep_delay))) %>%
  group_by(year, month, day) %>%
  summarise(
    cancelled_num = sum(cancelled),
    flights_num = n(),
  )

not_cancelled %>%
  group_by(arr_delay) %>%
  summarise(n = n())

ggplot(cancelled_per_day) +
  geom_point(aes(x = flights_num, y = cancelled_num))+

  ggplot(avg_delay)+
  geom_point(aes(x = delay, y = day))

cancelled_and_delays <- 
  flights %>%
  mutate(cancelled = (is.na(arr_delay) | is.na(dep_delay))) %>%
  group_by(year, month, day) %>%
  summarise(
    cancelled_prop = mean(cancelled),
    avg_dep_delay = mean(dep_delay, na.rm = TRUE),
    avg_arr_delay = mean(arr_delay, na.rm = TRUE)
  ) %>%
  ungroup()

ggplot(cancelled_and_delays) +
  geom_point(aes(x = avg_dep_delay, y = cancelled_prop))

#Yes, our intuition seems to hold to the notion that the amount of average cancellations
# increases as the average delay also increases.

#----------------------------------------------
# Question 3

# Part 1 answer:

flights %>%
  ggplot(aes(x=factor(hour), fill=dep_delay>0 | is.na(dep_delay)))+ 
  geom_bar()+
  coord_flip()

# Part 2 answer:

flights %>%
  ggplot(aes(x=factor(hour), fill=dep_delay<=dep_time))+ 
  geom_bar()+
  coord_flip()

# Based on the graph, 8:00 AM is the best time to book a flight if you want your
# departure time to be on time or early.


#----------------------------------------------
# Question 4

flights %>%
  ggplot(aes(x=factor(carrier), fill=dep_delay>=30))+ 
  geom_bar()+
  coord_flip()

# According the graph, the carrier that is most likely to experience a delay of
# 30 or more minutes is EV.

#----------------------------------------------
# Question 5

flights %>%
  ggplot(aes(x=factor(dest), fill=arr_delay<=1))+ 
  geom_bar()+
  coord_flip()

# According to the graph, ORD has the smallest average arrival delay, which also 
# plays to assumptions that could be made about its central location, size 
# (which means larger and faster planes) in comparison  to other destinations.