osm-cycle

This repository is home to a project looking at cycle path data on Open Street Map (OSM). The aims of the project are as follows:

1. Describe the cycle path data in OSM and test the hypothesis that it is a good source of data on cycle paths in Great Britain, suitable for academic research.

2. Compare the OSM cycle path data with alternative data sources.

3. Describe and explain the spatio-temporal distribution of additions to the the cycle path dataset and investigate how this corresponds with investment in cycle schemes overall.

As indicated by the above aims this is a data-driven project. So let's load some data:

Load Nottingham cycle data

Some example osm bicycle path data has been made available to get a handle on what it looks like. The bicycle paths were filtered from Postgres database using two methods: bicycle='yes', and highway = 'cycleway', to demonstrate that there are different ways to define a cycle path in OSM, as in the real world. This is how to load these layers into R:

library(rgdal)

## Loading required package: sp
## rgdal: version: 0.8-10, (SVN revision 478)
## Geospatial Data Abstraction Library extensions to R successfully loaded
## Loaded GDAL runtime: GDAL 1.10.0, released 2013/04/24
## Path to GDAL shared files: /usr/share/gdal/1.10
## Loaded PROJ.4 runtime: Rel. 4.8.0, 6 March 2012, [PJ_VERSION: 480]
## Path to PROJ.4 shared files: (autodetected)

pBikeYes <- readOGR("data/", "lines-cycle-yes")

## OGR data source with driver: ESRI Shapefile 
## Source: "data/", layer: "lines-cycle-yes"
## with 712 features and 66 fields
## Feature type: wkbLineString with 2 dimensions

pCycleway <- readOGR("data/", "lines-cycleway")

## OGR data source with driver: ESRI Shapefile 
## Source: "data/", layer: "lines-cycleway"
## with 969 features and 66 fields
## Feature type: wkbLineString with 2 dimensions

pBikeYes <- spTransform(pBikeYes, CRS("+init=epsg:27700"))
pCycleway <- spTransform(pCycleway, CRS("+init=epsg:27700"))

What are the attributes of these lines? Let's take a look:

plot(pBikeYes, col = "blue")
plot(pCycleway, col = "red", add = T)

Calclulate length

Let's look at the respective lengths of each line:

library(rgeos)

## rgeos version: 0.2-19, (SVN revision 394)
##  GEOS runtime version: 3.3.8-CAPI-1.7.8 
##  Polygon checking: TRUE

gLength(pBikeYes)/1000  # length in km

## [1] 109.1

gLength(pCycleway)/1000

## [1] 150.2

Clearly, paths marked with the cyclway attribute in the highway variable are slightly longer. Yet these points do not completely overlap, as illustrated in the map:

Preliminary time series analysis

osm2pgsql, used to create the datasets loaded above, does not import timestamp data by default. Therefore, we use a different dataset to illustrate time series data, this time loaded via the osmarR package. See my osm-tutorial GitHub page for more on this.

library(osmar)

## Loading required package: XML
## Loading required package: RCurl
## Loading required package: bitops
## Loading required package: gtools
## Loading required package: geosphere
## 
## Attaching package: 'osmar'
## 
## The following object is masked from 'package:utils':
## 
##     find

load("data/bike-paths-lds.RData")
plot(bikePaths)

class(bikePaths$nodes$attrs$timestamp)

## [1] "POSIXlt" "POSIXt"

tstamp <- as.POSIXct(bikePaths$nodes$attrs$timestamp, format = )
hist(bikePaths$nodes$attrs$timestamp, breaks = "year")

Further work

The above steps are very basic introductions to the osm datasets and how we we work with them. In this project we will make similar analyses, but on a much larger scale. We hope to generate policy relevant findings about a) the quality of cycle path data in OSM and b) the spatio-temporal distribution of paths, using imperfect OSM data as a proxy.