Geocoding and Reverse Geocoding Services are widely used to provide data about coordinate and location information, including longitude, latitude, formatted location name, administrative region with different levels. There are some packages can provide geocode service such as tidygeocoder, baidumap and baidugeo. However, some of them do not always provide precise information in China, and some of them are unavailable with the upgrade backend API.
amapGeocode is built to provide high precise geocoding and reverse
geocoding service, and it provides an interface for the AutoNavi(高德)
Maps API geocoding services. API docs can be found here and here. Here are
two main functions to use, one is getCoord() which needs a
character location name as an input, while the other one is
getLocation() which needs two numeric longitude and
latitude values as inputs.
The getCoord() function extracts coordinate information
from input character location name and outputs the results as
data.table, XML or
JSON (as list). And the getLocation() function
extracts location information from input numeric longitude and latitude
values and outputs the results as data.table,
XML or JSON (as list). With the
data.table format as output, it’s highly readable and can
be used as an alternative of data.frame
amapGeocode is inspired by baidumap and baidugeo. If you want to choose the Baidu Map API, these packages are good choices.
However, AutoNavi has significant high precise, in my case, the Results from Baidu were unsatisfactory.
planSince v0.5.1, parallel framework is implemented by furrr
package, of which backend is future package.
Refering to A
Future for R: Best Practices for Package Developers and
avoiding potential modification to the future strategy, we have removed
the automatically parallel operation from every function in
amapGeocode.
To turn on parallel operation support, just call
future::plan(multisession) # or any other future strategy.
Since v0.5, parallel operation finally comes to
amapGeocode with the parallel package as the
backend. There is a really huge performance improvement for batch
queries. And you are welcomed to make a benchmark by following
command.
library(amapGeocode)
library(future)
library(readr)
sample_site <-
read_csv("https://gist.githubusercontent.com/womeimingzi11/0fa3f4744f3ebc0f4484a52649f556e5/raw/47a69157f3e26c4d3bc993f3715b9ba88cda9d93/sample_site.csv")
str(sample_site)
# Here is the old implement
start_time <- proc.time()
old <- lapply(sample_site$address, amapGeocode:::getCoord.individual)
proc.time() - start_time
# Here is the new implement
plan(multisession)
start_time <- proc.time()
new <- getCoord(sample_site$address)
proc.time() - start_timeWhile parallel support is a totally threads depending operation, so you will get completely different speed on different devices.
You can install the released version of amapGeocode from CRAN with:
install.packages("amapGeocode")To install the development version, run following command:
remotes::install_github('womeimingzi11/amapGeocode')Before start geocoding and reverse geocoding, please apply a AutoNavi Map API Key. Set
amap_key globally by following command:
Then get results of geocoding, by getCoord function.
library(amapGeocode)
# An individual request
res <- getCoord("四川省中医院")
knitr::kable(res)| lng | lat | formatted_address | country | province | city | district | township | street | number | citycode | adcode |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 104.0431 | 30.6678 | 四川省成都市金牛区四川省中医院 | 中国 | 四川省 | 成都市 | 金牛区 | NA | NA | NA | 028 | 510106 |
# Batch requests
res <- getCoord(c("四川省中医院", "四川省人民医院", "成都中医药大学十二桥校区"))
knitr::kable(res)| lng | lat | formatted_address | country | province | city | district | township | street | number | citycode | adcode |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 104.0431 | 30.66780 | 四川省成都市金牛区四川省中医院 | 中国 | 四川省 | 成都市 | 金牛区 | NA | NA | NA | 028 | 510106 |
| 104.0390 | 30.66362 | 四川省成都市青羊区四川省人民医院 | 中国 | 四川省 | 成都市 | 青羊区 | NA | NA | NA | 028 | 510105 |
| 104.0439 | 30.66629 | 四川省成都市金牛区成都中医药大学十二桥校区 | 中国 | 四川省 | 成都市 | 金牛区 | NA | NA | NA | 028 | 510106 |
The responses we get from AutoNavi Map API is
JSON or XML. For readability, we
transform them to data.table,
by setting output argument as data.table by
default.
If you want to extract information from JSON or
XML. The results can further be parsed by
extractCoord.
# An individual request
res <- getCoord("成都中医药大学", output = "JSON")
res
#> $status
#> [1] "1"
#>
#> $info
#> [1] "OK"
#>
#> $infocode
#> [1] "10000"
#>
#> $count
#> [1] "1"
#>
#> $geocodes
#> $geocodes[[1]]
#> $geocodes[[1]]$formatted_address
#> [1] "四川省成都市金牛区成都中医药大学"
#>
#> $geocodes[[1]]$country
#> [1] "中国"
#>
#> $geocodes[[1]]$province
#> [1] "四川省"
#>
#> $geocodes[[1]]$citycode
#> [1] "028"
#>
#> $geocodes[[1]]$city
#> [1] "成都市"
#>
#> $geocodes[[1]]$district
#> [1] "金牛区"
#>
#> $geocodes[[1]]$township
#> list()
#>
#> $geocodes[[1]]$neighborhood
#> $geocodes[[1]]$neighborhood$name
#> list()
#>
#> $geocodes[[1]]$neighborhood$type
#> list()
#>
#>
#> $geocodes[[1]]$building
#> $geocodes[[1]]$building$name
#> list()
#>
#> $geocodes[[1]]$building$type
#> list()
#>
#>
#> $geocodes[[1]]$adcode
#> [1] "510106"
#>
#> $geocodes[[1]]$street
#> list()
#>
#> $geocodes[[1]]$number
#> list()
#>
#> $geocodes[[1]]$location
#> [1] "104.043284,30.666864"
#>
#> $geocodes[[1]]$level
#> [1] "兴趣点"extractCoord is created to get a result as a
data.table.
tb <- extractCoord(res)
knitr::kable(tb)| lng | lat | formatted_address | country | province | city | district | township | street | number | citycode | adcode |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 104.0433 | 30.66686 | 四川省成都市金牛区成都中医药大学 | 中国 | 四川省 | 成都市 | 金牛区 | NA | NA | NA | 028 | 510106 |
get results of reverse geocoding, by getLocation
function.
res <- getLocation(104.043284, 30.666864)
knitr::kable(res)| formatted_address | country | province | city | district | township | citycode | towncode |
|---|---|---|---|---|---|---|---|
| 四川省成都市金牛区西安路街道成都中医药大学附属医院腹泻门诊成都中医药大学十二桥校区 | 中国 | 四川省 | 成都市 | 金牛区 | 西安路街道 | 028 | 510106024000 |
extractLocation is created to get a result as a
data.table.
get results of reverse geocoding, by getAdmin
function.
There is a difference between getAdmin and other function, no matter
the output argument is data.table or not, the
result won’t be a jointed table by different parent administrative
region. For example, with the output = data.table, all the
lower level administrative region of Province A and Province B will be
bound as one data.table, respectively. But the table of province A and
table of province B won’t be bound further.
Because this function supports different administrative region levels, it is nonsense to bind their results.
res <- getAdmin(c("四川省", "成都市", "济宁市"))
knitr::kable(res)
|
|
|
extractAdmin is created to get results as tibble.
get results of reverse geocoding, by convertCoord
function, here is how to convert coordinate from gps to AutoNavi.
Please not, this is still a very experimental function because I have no experience at converting coordinates. The implementation of this input method is not as delicate as I expect. If you have any good idea, please let me know or just fork repo and pull a reques.
res <- convertCoord("116.481499,39.990475", coordsys = "gps")
knitr::kable(res)| lng | lat |
|---|---|
| 116.4876 | 39.99175 |
extractConvertCoord is created to get result as
data.table.
It’s very common for API upgrades to make the downstream application, like amapGeocode,which is unavailable. Feel free to let me know once it’s broken or just open an Issue.
Hex Sticker was created by hexSticker package with the world data from rnaturalearth.
Please note that the amapGeocode project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.