param-repro-rsch.Rmd

---
title: "Parameterizable Reproducible Research"
author: "Your Name Here"
output: html_document
---

`r opts_chunk$set(echo=FALSE, results='asis', cache=TRUE, message = FALSE, warning = FALSE)`

This document illustrates the results of a parameterizable reproducible research process. It pulls
train delay data from [WMATA](http://wmata.com/), then does some simple analyses on that data.

```{r Load_Data}

wmata_url <- "https://docs.google.com/spreadsheet/ccc?key=0ApyDIPnioxR3dEswVzIxb2NCdnZTLUZDZE9pTnhMcGc&usp=sharing&output=csv"
wmata_dat <- content(GET(wmata_url))

```

`r nrow(wmata_dat)` delays were loaded. 

```{r Process_Data}
wmata_dat <- wmata_dat %>%
  filter(Line == line_color) %>%
  mutate(Date = parse_date_time(paste(Date, str_replace_all(Time, "\\.", "")),
                                "%m/%d/%Y %I/%M %p"))
```

The data was filtered to only include the `r nrow(wmata_dat)` events on the `r line_color` line.
The first delay on that line was at
`r format(min(wmata_dat$Date, na.rm=TRUE))`; the last was at `r format(max(wmata_dat$Date, na.rm=TRUE))`.
Here are the three most common causes for delays on that line:
_`r paste(names(head(sort(table(wmata_dat$Cause), decreasing = TRUE), 3)), sep=", ")`_.

This table shows the mean delay and counts of the most frequent causes:

```{r Summarize_Data}

delay_sum <- wmata_dat %>% 
  group_by(Cause) %>% 
  summarise(mean_delay=mean(Delay, na.rm=TRUE), n=n()) %>% 
  arrange(-n, -mean_delay) %>% 
  head(10)

pander(delay_sum)

```

And this graph shows when delays happened by date and hour.

```{r Plot_Data, fig.height=4, fig.width=6}
ggplot(wmata_dat, aes(Date, hour(Date), size=Delay)) + 
  geom_jitter() +
  scale_y_continuous("Hour", breaks=c(0, 8, 12, 17, 24))

```