Week 5 Assignment

Introduction to the Data

The data set we are starting with is one on the arrival time of 2 different airlines at 5 different destinations and for each of these airline and destination combinations, the number of flights that arrived on time vs delayed. What we will be looking into is the percent of flights delayed to each location for each airline, and which airline would be best to take for each destination.

destinations <- as.data.frame(t(arrivals[1, !is.na(unlist(arrivals[1, ]))]))
carriers <- arrivals$X1[!is.na(arrivals$X1)]
arrivals <- arrivals[-1, -c(1,2)]

Starting off I want to get all the arrival destinations, and all the carriers. After that I can remove those rows form the data frame.

carrier_flights <- data.frame(on_time = integer(0), delayed = integer(0), stringsAsFactors = FALSE)
destinations_full <- destinations
carriers_full <- data.frame(carrier = character(length(arrivals) * length(carriers)), stringsAsFactors = FALSE)
i = 1
j = 1

Here I am initializing variables to contain all carrier, destination, and arrival delay combinations.

while (i < nrow(arrivals)) {
carrier_flights <- rbind(carrier_flights, as.data.frame(t(arrivals[i:(i+1), ])))
carriers_full$carrier[j:(j+length(arrivals)-1)] <- carriers[ceiling(j/length(arrivals))]
if(i > 1){
  destinations_full <- rbind(destinations_full, destinations)
}
j <- j + length(arrivals)
i <- i+3
}
df_full <- cbind(carriers_full, destinations_full, carrier_flights)
colnames(df_full) <- c('carrier', 'destination', 'on_time', 'delayed')
rownames(df_full) <- NULL
df_full[, c("on_time", "delayed")] <- lapply(df_full[, c("on_time", "delayed")], as.integer)
print(df_full)

##    carrier   destination on_time delayed
## 1   ALASKA   Los Angeles     497      62
## 2   ALASKA       Pheonix     221      12
## 3   ALASKA     San Diego     212      20
## 4   ALASKA San Francisco     503     102
## 5   ALASKA       Seattle    1841     305
## 6  AM WEST   Los Angeles     694     117
## 7  AM WEST       Pheonix    4840     415
## 8  AM WEST     San Diego     383      65
## 9  AM WEST San Francisco     320     129
## 10 AM WEST       Seattle     201      61

In this loop I treat the data set as if there could be any number of carriers or arrival destinations added for future proofing. What I did is loop through and for each carrier I transposed the on time/delayed numbers to get them in columns, as well as added a full set of rows of the corresponding destinations and carriers so that they can be combined to form one large data set.

df_summed_carrier <- df_full %>% group_by(carrier) %>%   summarise(
    on_time = sum(on_time),
    delayed = sum(delayed)
  )
df_summed_carrier <- df_summed_carrier %>% mutate(delay_percent = (delayed / (on_time + delayed)) * 100)
ggplot(df_summed_carrier, aes(x = carrier)) +
  geom_bar(aes(y = on_time, fill = 'On Time'), stat = 'identity', position = 'stack') +
  geom_bar(aes(y = delayed, fill = 'Delayed'), stat = 'identity', position = 'stack') +
  scale_fill_manual(values = c('On Time' = 'green', 'Delayed' = 'red')) +
  labs(x = "Carrier", y = "Total Flights", title = "Total On-Time and Delayed Flights per Carrier") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

print(df_summed_carrier[,c('carrier', 'delay_percent')])

## # A tibble: 2 × 2
##   carrier delay_percent
##   <chr>           <dbl>
## 1 ALASKA           13.3
## 2 AM WEST          10.9

Here we can see that AM WEST has significantly more flights than ALASKA, and to be expected with that, they have more delayed flights as well, but a lower overall delay percentage at 10.9% vs 13.3% for ALASKA.

df_full$delay_percent <- (df_full$delayed / (df_full$on_time+df_full$delayed))*100.0
ggplot(df_full, aes(x = destination, y = delay_percent, fill = carrier)) +
  geom_bar(stat = "identity", position = "dodge") +
  labs(x = "Destinations", y = "Percent Delayed", fill = "Carrier") +
  ggtitle("Percent of Flights Delayed by Destination and Carrier") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

destinations_split <- split(df_full, df_full$destination)
min_delay_rows <- lapply(destinations_split, function(x) x[which.min(x$delay_percent), ])
df_low_delay <- result_df <- do.call(rbind, min_delay_rows)
rownames(df_low_delay) <- NULL
print(df_low_delay[,c("destination", "carrier")])

##     destination carrier
## 1   Los Angeles  ALASKA
## 2       Pheonix  ALASKA
## 3     San Diego  ALASKA
## 4 San Francisco  ALASKA
## 5       Seattle  ALASKA

In this final section I find the carrier with the lowest percent of flights delayed for each destination in a way where if there were more carriers or destinations added to the csv then they would automatically flow through to this final table.

Week 5 Assignment

Brandon Cunningham

2024-02-20

Introduction to the Data

Conculsions