Hubway Network Analysis

---
title: "Hubway Network Excercise"
author: "Brian Gridley"
date: "February 19, 2018"
output: pdf_document
---

```{r}
# load the two datasets
hubedge <- read.csv("data/hubway [Edges].csv", header = TRUE)
hubnodes <- read.csv("data/hubway [Nodes].csv", header = TRUE)

hubedge
# could look at just one source or look at trips on a certain date or look at certain age group or gender
arrange(hubnodes, Id)

library(igraph)

# filter it out to look at all routes going in and out of Union Sq... interested in this bc it is an area where a lot of people live but there is currently no train access, so the only other public transit option is the bus.. want to see how active it is

# filter it to look at one from each lettered section in terminal name
 # 7 = fan pier
 # 52 = newbury st
 # 27 = roxbury crossing
 # 37 = newbalance
 # 69 = coolidge corner
 # 90 = lechmere
 # 78 = Union Sq

filter(hubedge, Source == "7" | Source == "52" | Source == "27" | Source == "37" | Source == "69" | Source == "90" | Source == "78")
#534 rows
filter(hubedge, Target == "7" | Target == "52" | Target == "27" | Target == "37" | Target == "69" | Target == "90" | Target == "78")
#535
filter(hubedge, Source == "78" | Target == "78")
#138

citywide <- filter(hubedge, Source == "7" | Source == "52" | Source == "27" | Source == "37" | Source == "69" | Source == "90" | Source == "78" | Target == "7" | Target == "52" | Target == "27" | Target == "37" | Target == "69" | Target == "90" | Target == "78")

# too big, take a random sample from this... 50 rows
citywidesample <- sample_n(citywide, 50)

citywidelarge <- sample_n(hubedge, 50)

# look at one workday morning... random day... tuesday, August 21, between 5 and 9 am
august_day <- filter(hubedge, substr(hubedge$start_date, 1,10) == "2012-08-21" & substr(hubedge$end_date, 12,16) < "09:00" & substr(hubedge$start_date, 12,16) > "07:00")

arrange(august_day, start_date)
#48 rows
```

```{r}
aug <- graph.data.frame(august_day)   # parse the dataframe into the igraph object mis



E(aug)
V(aug)
V(aug)$name
vertex_attr(aug)



# Set edge width based on duration:
E(aug)$width <- E(aug)$duration/500

name <- V(aug)$name 

# graph it
plot(aug, vertex.label=name, vertex.size = 10, edge.arrow.size=.2, vertex.color="gold", 
     vertex.frame.color="navy", vertex.label.color="navy", 
     vertex.label.cex=0.9,edge.curved=0.2, edge.color = "gray25")                         # render the igraph object



```


```{r}

# now lets try an interactive network plot
# https://christophergandrud.github.io/networkD3/
install.packages("networkD3")
library(networkD3)

# for the d3 network we need the original data frame with the edge list
#lesmis <- read.delim("data/lesmis.txt", header = FALSE)
n <- simpleNetwork(august_day)
n
saveNetwork(n, file = "OneAugustMorning.html") #save our javascript network to a html file

```