---
output: pdf_document
---
Airline Delays in the First Course (part 4: analyzing the data from Flagstaff)
========================================================
Nicholas Horton, nhorton@amherst.edu, July 15, 2014
------------------------------------------------------

#### Accessing the data


```{r, message=FALSE, tidy=FALSE}
require(mosaic)
options(digits=3)
trellis.par.set(theme=col.mosaic())  # get a better color scheme for lattice
load("Meetup-talk/ICOTS-FLG.rda")
names(ds)
filter(ds, Year==2013 & DayofMonth==20)  %>%  # Saturday, July 20th, 2013
  select(DayofMonth, Month, Year, Origin, Dest, CRSDepTime, ArrDelay, Cancelled) %>%
  arrange(CRSDepTime)
ds = mutate(ds, TimeOfDay = cut(CRSDepTime, breaks=c(0, 1200, 1800, 2400), 
    labels=c("morning", "afternoon", "evening")))
ds = mutate(ds, delayorcancel = ifelse(is.na(ArrDelay) | ArrDelay > 15, "yes", "no"))
```

```{r}
favstats(~ CRSDepTime, data=ds)
tally(~ TimeOfDay, data=ds)
```

```{r}
tally(~ Cancelled, format="percent", data=ds)
tally(~ delayorcancel, format="percent", data=ds)

tally(~ Cancelled | TimeOfDay, format="percent", data=ds)
tally(~ delayorcancel | TimeOfDay, format="percent", data=ds)

```

```{r}
favstats(~ ArrDelay, data=ds)
ds = mutate(ds, ActDelay = ifelse(ArrDelay < 0, 0, ArrDelay))
favstats(~ ActDelay, data=ds)
favstats(ActDelay ~ TimeOfDay, data=ds)

bwplot(ActDelay ~ TimeOfDay, ylim=c(-10, 120), main="July flights from Flagstaff, 2011-2013", ylab="Actual arrival delay (in minutes)", data=ds)
bwplot(TimeOfDay ~ ActDelay, xlim=c(-10, 120), main="July flights from Flagstaff, 2011-2013", xlab="Actual arrival delay (in minutes)", data=ds)
```



```{r}
densityplot(~ ArrDelay, groups=TimeOfDay, auto.key=TRUE, 
            xlab="Arrival delay (in minutes)", xlim=c(-35, 120), data=ds)
densityplot(~ ArrDelay, groups=TimeOfDay, adjust=1/2, auto.key=TRUE, 
            xlab="Arrival delay (in minutes)", xlim=c(-35, 120), data=ds)
```

```{r}
ds2 = filter(ds, Year==2013)
sort(tally(~ TailNum, data=ds2), decreasing = TRUE)