| ##################################################################################### | |
| # Title: Nigeria Urban Mask | |
| # Date: Sep 2015 | |
| # Project: HarvestChoice | |
| # Author: Bacou, Melanie <mel@mbacou.com> | |
| ##################################################################################### | |
| library(raster) | |
| library(tmap) | |
| setwd("~/Projects/hc-cell5m") |
| # Example #1 (takes 2 calls to request a download file) | |
| # Returns wheat harvested area by district in Ghana in STATA format | |
| $ curl http://hcapi.harvestchoice.org/ocpu/library/hcapi3/R/hcapi3 -d "{var: 'whea_h', iso3: 'GHA', by: 'ADM2_NAME_ALT', format: 'dta'}" -X POST | |
| /ocpu/tmp/x0a3355977c/R/.val | |
| /ocpu/tmp/x0a3355977c/messages | |
| /ocpu/tmp/x0a3355977c/stdout | |
| /ocpu/tmp/x0a3355977c/source | |
| /ocpu/tmp/x0a3355977c/console | |
| /ocpu/tmp/x0a3355977c/info | |
| /ocpu/tmp/x0a3355977c/files/DESCRIPTION |
| ## Spatial Targeting Tool | |
| # Summarize wheat area and household wealth index over AEZ class "Tropic - warm / semiarid", | |
| # pop density between 60-100 pp/sq.km. and low market access (under 5hrs) | |
| $ time curl http://hcapi.harvestchoice.org/ocpu/library/hcapi3/R/reClassify/json -d '{"var" : ["whea_h", "wealth"], "by" : {"AEZ16_CLAS": "Tropic - warm / semiarid", "PD05_RUR" : [60, 100], "TT_20K" : [0, 5]}}' -X POST -H "Content-Type:application/json" | |
| [ | |
| { | |
| "TT_20K": "(0,5]", | |
| "whea_h": 111336, | |
| "wealth": 2 | |
| }, |
| ## Improved timings of OpenCPU stack using persistent Rserve session | |
| # Summary by district | |
| $ time curl http://127.0.0.1/ocpu/library/hcapi3/R/getLayer \ | |
| -d '{"var" : ["whea_h", "AEZ16_CLAS"], "by" : "ADM2_NAME_ALT"}' \ | |
| -X POST -H "Content-Type:application/json" | |
| /ocpu/tmp/x010a92e97a/R/.val | |
| /ocpu/tmp/x010a92e97a/source | |
| /ocpu/tmp/x010a92e97a/console |
| ## Comparing execution times between OpenCPU (RApache) and FastRWeb (Rserve and HTTP). | |
| ## Same requests and same code. Note that FastRWeb is up to 13 times faster on the 2nd call. | |
| # OpenCPU: summarize 2 layers across districts for the whole of SSA, return json of summary table | |
| # and then json of all pixel values for mapping | |
| $ time curl http://127.0.0.1/ocpu/library/hcapi3/R/getLayer/json \ | |
| -d '{"var" : ["whea_h", "AEZ16_CLAS"], "by" : "ADM2_NAME_ALT"}' -X POST -H "Content-Type:application/json" | |
| [...] | |
| { |
| # Voila | |
| require(raster) | |
| can <- getData("GADM", country="CAN", level=0) | |
| # And then simplify boundaries for plotting | |
| can <- gSimplify(can, tol=0.01, topologyPreserve=TRUE) | |
| # Plot raster with full control | |
| par(oma=c(0,0,0,4)) | |
| plot(r, col=cc, xlim=xlim, ylim=ylim, main=NULL, legend.width=1.2, |
| # Using ggmap to make more complex maps with inset, etc.. | |
| ethMap <- qmap("Ethiopia", zoom=6, color="color", source="google", maptype="terrain", | |
| extent="panel", padding=0.02) | |
| addisMap <- qmap("Addis Ababa", zoom=12, color="color", source="google", maptype="terrain", | |
| extent="panel", padding=0) | |
| layer.dhs <- geom_point(aes(x=lon, y=lat), data=loc.eth@data, | |
| size=1.5, colour="blue", fill="blue", shape=2) |
tza.stata <- sp2tmap(tza) names(tza.stata) <- c(“ID”, “X”, “Y”)
tza.stata <- data.table(tza.stata) tza@data <- data.table(tza@data) tza@data[, rn := 1:dim(tza@data)[1]] setkey(tza.stata, ID) setkey(tza@data, rn) tza.stata$ADM2_CODE <- tza.stata[tza@data][, ADM2_CODE]