-
Notifications
You must be signed in to change notification settings - Fork 9
/
README.Rmd
136 lines (99 loc) · 8.03 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r setup, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
library(areal)
library(dplyr)
library(sf)
data(ar_stl_asthma, package = "areal")
asthma <- ar_stl_asthma
data(ar_stl_race, package = "areal")
race <- ar_stl_race
data(ar_stl_wards, package = "areal")
wards <- ar_stl_wards
```
# areal <img src="man/figures/arealLogo.png" align="right" />
[](https://github.com/chris-prener/areal/actions)
[](https://codecov.io/github/chris-prener/areal?branch=main)
[](https://cran.r-project.org/package=areal)
[](https://cran.r-project.org/web/checks/check_results_areal.html)
[](https://www.r-pkg.org/pkg/areal)
[](https://zenodo.org/badge/latestdoi/152279647)
[](https://doi.org/10.21105/joss.01221)
Areal interpolation is the process making estimates from a source set of polygons to an overlapping but incongruent set of target polygons. One challenge with areal interpolation is that, while the processes themselves are well documented in the academic literature, implementing them often involves "reinventing the wheel" by re-creating the process in the analyst's tool choice.
While the `R` package `sf` does offer a basic interface for areal weighted interpolation (`st_interpolate_aw`), it lacks some features that we use in our work. The `areal` package contains a suite tools for validation and estimation, providing a full-featured workflow that fits into both modern data management (e.g. `tidyverse`) and spatial data (e.g. `sf`) frameworks.
### *Joural of Open Souce Software* Article
An [article](https://joss.theoj.org/papers/10.21105/joss.01221) describing `areal`'s approach to areal weighted interpolation has been published in the [*The Journal of Open Source Software*](https://joss.theoj.org/). The article includes benchmarking of `areal` performance on several data sets. Please [cite the paper](/inst/CITATION) if you use `areal` in your work!
## Installation
The easiest way to get `areal` is to install it from CRAN:
``` r
install.packages("areal")
```
The development version of `areal` can be accessed from GitHub with `remotes`:
```r
# install.packages("remotes")
remotes::install_github("chris-prener/areal")
```
Note that installations that require `sf` to be built from *source* will require additional software regardless of operating system. You should check the [`sf` package website](https://r-spatial.github.io/sf/) for the latest details on installing dependencies for that package. Instructions vary significantly by operating system.
## Usage
Two function prefixes are used in `areal` to allow users to take advantage of RStudio's auto complete functionality:
* `ar_` - data and functions that are used for multiple interpolation methods
* `aw_` - functions that are used specifically for areal weighted interpolation
### Data
The package contains four overlapping data sets:
* `ar_stl_race` (2017 ACS demographic counts at the census tract level; *n* = 106)
* `ar_stl_asthma` (2017 asthma rates at the census tract level; *n* = 106)
* `ar_stl_wards` (the 2010 political subdivisions in St. Louis; *n* = 28).
* `ar_stl_wardsClipped` (the 2010 political subdivisions in St. Louis clipped to the Mississippi River shoreline; *n* = 28).
These can be used to illustrate the core functionality of the package. The following examples assume:
```r
> library(areal)
>
> race <- ar_stl_race
> asthma <- ar_stl_asthma
> wards <- ar_stl_wards
```
### Areal Weighted Interpolation
`areal` currently implements an approach to interpolation known as areal weighted interpolation. It is arguably the simplest and most common approach to areal interpolation, though it does have some drawbacks (see the [areal weighted interpolation vignette](https://chris-prener.github.io/areal/articles/areal-weighted-interpolation.html) for details). The basic usage of `areal` is through the `aw_interpolate()` function. This is a pipe-able function that allows for the simultaneous interpolation of multiple values.
In this first example, the total estimated population (`TOTAL_E`) of each ward is calculated from its overlapping census tracts:
```{r iteration}
aw_interpolate(wards, tid = WARD, source = race, sid = "GEOID",
weight = "sum", output = "sf", extensive = "TOTAL_E")
```
This example outputs a simple features (`sf`) object and uses one of two options for calculating weights. All of these arguments are documented both within the package (use `?aw_interpolate`) and on the [package's website](https://chris-prener.github.io/areal/).
What results from `aw_interpolate()` is mapped below. Total population per census tract in St. Louis is mapped on the left in panel A. Using `aw_interpolate()` as we did in the previous example, we estimate population counts for Wards in St. Louis from those census tract values. These estimated values are mapped on the right in panel B.
```{r exampleMap, echo=FALSE, out.width = '100%'}
knitr::include_graphics("man/figures/exampleMap.png")
```
Both extensive and intensive data can be interpolated simultaneously by using both the `extensive` and `intensive` arguments. In this second example, the asthma and race data are combined, and estimates for both the population values and asthma rates are calculated for each ward from its overlapping census tracts:
```{r mixed}
# remove sf geometry
st_geometry(race) <- NULL
# create combined data
race %>%
select(GEOID, TOTAL_E, WHITE_E, BLACK_E) %>%
left_join(asthma, ., by = "GEOID") -> combinedData
# interpolate
wards %>%
select(-OBJECTID, -AREA) %>%
aw_interpolate(tid = WARD, source = combinedData, sid = "GEOID",
weight = "total", output = "tibble",
extensive = c("TOTAL_E", "WHITE_E", "BLACK_E"),
intensive = "ASTHMA")
```
Another advantage of `areal` is that the interpolation process is not a "black box", but rather can be manually completed if necessary. Functions for validating data, previewing the areal weights, and walking step-by-step through the interpolation process are provided. See the [areal weighted interpolation vignette](https://chris-prener.github.io/areal/articles/areal-weighted-interpolation.html) for additional details about this workflow.
## Road-map
We are planning to experiment with at least three additional techniques for areal interpolation for possible inclusion into the package. These include:
- [Pycnophylactic method](https://github.com/chris-prener/areal/issues/1) (raster based, eliminates the sharp transitions in value between target features)
- [Binary dasymetric method](https://github.com/chris-prener/areal/issues/2) (incorporates ancillary data so that population is not assumed to be evenly distributed within units)
- [3-class regression dasymetric method](https://github.com/chris-prener/areal/issues/3) (allows for a more complex estimation based on multiple forms of ancillary data)
We do not have a timeline for these experiments, though we are planning to begin experimenting with the pycnophylactic method in the coming months. We will be keeping the issues (linked to above) updated with progress. If you are interested in bringing these techniques to `R`, please feel free to contribute to the development of `areal`. The best place to start is bt checking in on our GitHub issues for each technique to see what help is needed!
## Contributor Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](https://chris-prener.github.io/areal/CODE_OF_CONDUCT.html). By participating in this project you agree to abide by its terms.