BiostatR Blog

You can find many information in Wikipedia pages but sometimes in can be long to gather them manually. R can be used as a sucker of information. Let see how it can retrieve the coordinates of volcanos in Central America. The web pages are here: https://en.wikipedia.org/wiki/List_of_volcanoes_in_Mexico https://en.wikipedia.org/wiki/List_of_volcanoes_in_Guatemala https://en.wikipedia.org/wiki/List_of_volcanoes_in_El_Salvador https://en.wikipedia.org/wiki/List_of_volcanoes_in_Nicaragua https://en.wikipedia.org/wiki/List_of_volcanoes_in_Costa_Rica library(maps) library(mapdata) library(maptools) library(XML) # Note a problem in readHTMLTable. It cannot be used directly with url: # readHTMLTable(url, header=FALSE, stringsAsFactors = FALSE) generates an error # load a file in temporary directory url <- "https://en.wikipedia.org/wiki/List_of_volcanoes_in_Mexico" dest <- paste(tempdir(), "/List_of_volcanoes_in_Mexico.html", sep="")

Imagine a logistic model with these data: t <- c(25, 25, 26, 28, 30, 32, 33, 33, 34) n <- c(10, 12, 13, 15, 10, 16, 10, 12, 15) x <- c(10, 12, 13, 15, 0, 0, 0, 0, 0) S <- -0.5 P <- 29 Lbinom <- function(data, x) -sum(dbinom(x=data$x, size=data$size, prob=1/(1+exp((1/x["S"])*(x["P"]-data$t))), log=TRUE)) Lbinom(data=list(x=x, size=n, t=t), x=c(P=P, S=S)) library("HelpersMG") parameters_mcmc <- data.frame(Density=c('dunif', 'dunif'),              Prior1=c(25, -2), Prior2=c(32, 2), SDProp=c(0.35, 0.2),              Min=c(25, -2), Max=c(32, 2), Init=c(29, -0.5), stringsAsFactors = FALSE,              row.names=c('P', 'S')) mcmc_run <- MHalgoGen(n.iter=50000, parameters=parameters_mcmc, data=list(x=x, size=n, t=t),                        likelihood=Lbinom, n.chains=1, n.adapt=100, thin=1, trace=1, adaptive = TRUE) plot(mcmc_run, parameters = 1, xlim=c(20, 35)) plot(mcmc_run, paramet

Let calculate of sample of 100000 points defining credibility interval from MCMC with Metropolis-Hasting algorithm based on a logistic model: t <- c(25, 25, 26, 28, 30, 32, 33, 33, 34) n <- c(10, 12, 13, 15, 10, 16, 10, 12, 15) x <- c(10, 11, 12, 13, 5, 3, 1, 0, 0) S <- -0.5 P <- 29 Lbinom <- function(data, x) -sum(dbinom(x=data$x, size=data$size, prob=1/(1+exp((1/x["S"])*(x["P"]-data$t))), log=TRUE)) Lbinom(data=list(x=x, size=n, t=t), x=c(P=P, S=S)) library("HelpersMG") parameters_mcmc <- data.frame(Density=c('dunif', 'dunif'),              Prior1=c(25, -2), Prior2=c(32, 2), SDProp=c(0.35, 0.2),              Min=c(25, -2), Max=c(32, 2), Init=c(29, -0.5), stringsAsFactors = FALSE,              row.names=c('P', 'S')) mcmc_run <- MHalgoGen(n.iter=50000, parameters=parameters_mcmc, data=list(x=x, size=n, t=t),                        likelihood=Lbinom, n.chains=1, n.adapt=100, thin=1

First you must install gdal library: Load  GDAL Complete  at this link. Use the two installers present in that folder. Then enter in terminal: # this is optional if you have done it already. Check with # cat  ~/.profile sudo echo  'export PATH=/Library/Frameworks/GDAL.framework/Programs:$PATH' >> ~/.profile export PATH=/Library/Frameworks/GDAL.framework/Programs:$PATH cd ~/Downloads/ curl -O   https://cran.r-project.org/src/contrib/rgeos_0.3-23.tar.gz R CMD INSTALL  ~/Downloads/ rgeos_0.3-23.tar.gz --configure-args='--with-geos-config=/Library/Frameworks/GEOS.framework/unix/bin/geos-config'