BiostatR Blog

Analyse de la sensibilité d'un automate cellulaire # Fonction qui calcule le nombre de cases occupées (=1) en fonction d'une variable X # qui est un data.frame qui contient les probabilités de mourir à chaque pas de temps (p) et  # le nombre de cases (/9) pour obtenir un décès d'étouffement. # La variable t donne les temps pour lesquels on retourne la taille de la population # Si t est une valeur, on retourne juste la valeur pour ce temps, si t est un vecteur,  # on retourne toutes les valeurs pour les temps t ac <- function(X, t=t) {   Nfinal <- NULL   for (grandtotal in 1:nrow(X)) {     penc <- X[grandtotal, 1]     v <- X[grandtotal, 2]*9     m <- matrix(data = 0, ncol=12, nrow=12)   for (i in 1:10) {     col <- sample(1:10, 1)+1     row <- sample(1:10, 1)+1     m[row, col] <- 1   }   N <- NULL   for (tps in 1:max(t)) {   ...

Everyone has done a linear regression... but not everyone knows really how to do a linear regression. There are several methods which produce different outputs. Here is a little synthesis: # Let generate data. Note the the scale for x and y are different. # the set.seed(1) makes sure that the data are always the same # to make reproduced examples set.seed(1) x <- 1:10 y <- 10*x+rnorm(10, 5, 20) plot(x, y, bty="n", las=1) # First method. Only error on y axis is supposed # The square of the residuals on y axis is minimized # Why the sum of square? It is grounded on the likelihood of Gaussian distribution for residuals lm(y~ x) abline(lm(y~ x), col="red") # second, very similar to previous one, only error on x axis is supposed # The square of the residuals on x axis is minimized ir <- coef(lm(x~y)) print(c(-ir[1]/ir[2],1/ir[2])) abline(-ir[1]/ir[2],1/ir[2], col="blue") # The choice between error inputed...

grib and grib2 are format used to store meteorological data. It is the native format used by ECMWF server. To read it in R with MacOSX: Using Macport, install the library wgrib2: sudo port install wgrib2 You must also install the wgrib library: Look at here: https://rda.ucar.edu/datasets/ds083.2/software/wgrib_install_guide.txt Then you can install rNOMAD package in R You must install gdal library You can use also the raster package with rgdal and goal library 4 packages can be used: rNOMAD, rgdal, raster, stars I found that the simple is to use rgdal

After install, open Xcode and install commande line tools at front. In terminal: sudo xcodebuild -license type agree after many pages then sudo xcode-select --install defaults write org.R-project.R force.LANG fr_FR.UTF-8 sudo installer -pkg /Library/Developer/CommandLineTools/Packages/macOS_SDK_headers_for_macOS_10.14.pkg -target /

This note is based on a discussion with Jean-Michel Guillon. The idea was to test an idea from this this paper: Sterne, J.A.C., Smith, G.D., 2001. Sifting the evidence-what’s wrong with significance tests? British Medical Journal 322, 226-231. If our prior opinion about the risk ratio is vague (we consider a wide range of values to be equally likely) then the results of a frequentist analysis are similar to the results of a bayesian analysis; both are based on what statisticians call the likelihood for the data: • The 95% confidence interval is the same as the 95% credible interval, except that the latter has the meaning often incorrectly ascribed to a confidence interval; • The (one sided) P value is the same as the bayesian posterior probability that the drug increases the risk of death (assuming that we found a protective effect of the drug). The two approaches, however, will give different results if our prior opinion is not vague, relative to the amount of informatio...

I had some difficulties to use the OR | operator in regular expressions. Let do some exercises and examine the results: gregexpr("bc", "abcd") : return 2; normal, bc is located at the second position gregexpr("b|c", "abcd") : return 2 3; | is the OR operator and it will return the position of b OR c this is the same than : gregexpr("[bc]", "abcd") or gregexpr("[b|c]", "abcd") But | and [] are not always the same: Let do more complicated: gregexpr("ab|cd", "abcd") My problem was about the priority order of | operator. The return here is 1 3 It means that the search was for a followed by b OR c followed by d. Let do more complicated : gregexpr("a[bc]d", "abcd") : return -1; it search for abd or acd. None exist In practice, it is good to remember that | separates blocks of comparisons.

If you want search for a number using grep, you have 4 options: [0123456789] or [0-9] or \d or [:digit:] Here is a comparison: library(microbenchmark) microbenchmark({grep(pattern="[0123456789][0123456789][0123456789][0123456789]", "Ussiosuus8980JJDUD98")},                 {grep(pattern="[0-9][0-9][0-9][0-9]", "Ussiosuus8980JJDUD98")},                 {grep(pattern="[[:digit:]][[:digit:]][[:digit:]][[:digit:]]", "Ussiosuus8980JJDUD98")},                 {grep(pattern="\\d\\d\\d\\d", "Ussiosuus8980JJDUD98")}, times = 1000L)     min      lq      mean  median      uq    max neval  cld  21.729 22.1840 22.887585 22.3090 22.4645 64.542  1000    d   5.219  5.3740  5.696937  5.4665  5.8495 47.618  1000 a     ...