RafaelSdeSouza/CosmoR

title	author	date	output
Generalized Linear Models for Astronomy	Rafael S. de Souza.	August 14th, 2014	pdf_document

Installation

CosmoGLM is available from github. The current development version can be installed using devtools.

library(devtools)
install_github("CosmoR", username="RafaelSdeSouza", subdir="CosmoGLM")

This is an R Markdown document to explain in simple terms the use of Generalized Linear Models into Astronomy reproducing the main results from Rafael S. de Souza et al (2014).

Reading the data

Set working directory to the data folder (replace by your own directory)

data_path<-"/Users/rafael/Dropbox/artigos/Meusartigos/IAA-WGC/GLMs/Simulation/data/"

Biffi_data<-read.table(file=paste(data_path,"Biffi2014.csv",sep=""),
                       header=TRUE)

Biffi_original<-Biffi_data

Problem 1: xmol, Z, SFR. SFR is the response variable

Biffi_data<-Biffi_data[,c("SFR","Xmol","Z")]

Transforming variable into numeric (required by GLM packages)

Biffi_data$SFR[which(Biffi_data$SFR==0)]<-0
Biffi_data$SFR[which(Biffi_data$SFR>0)]<-1
Biffi_data$SFR<-as.numeric(Biffi_data$SFR)
Biffi_data$Z<-scale(Biffi_data$Z)#Scaling and centering
Biffi_data$Xmol<-scale(Biffi_data$Xmol)#Scaling and centering

Frequentist

Fglm <-glm(SFR~Z+Xmol,family=binomial(link="logit"),
                 data = Biffi_data)

plotProb(Fglm)+ylab("Predicted Probabilities for SF")+xlab("")+
ggtitle("Frequentist Logistic Regression")+coord_cartesian(ylim = c(0,1.05))