This book is in Open Review. I want your feedback to make the book better for you and other readers. To add your annotation, select some text and then click the on the pop-up menu. To see the annotations of others, click the in the upper right hand corner of the page

Cours 26 Version seulment corr >0.5

# inspired by the function of Catherine Williams
onlySignCorr <- function(df, file){
  corr <- cor(df)
  
  #drop perfect correlations
  corr[corr == 1] <- NA 
  #turn into a 3-column table
  corr <- as.data.frame(as.table(corr))
  #remove the NA values from above 
  corr <- na.omit(corr) 
  
  #select significant values  
  corr <- subset(corr, abs(Freq) > 0.5) 
  #sort by highest correlation
  corr <- corr[order(-abs(corr$Freq)),] 
  #turn corr back into matrix in order to plot with corrplot
  mtx_corr <- reshape2::acast(corr, Var1~Var2, value.var="Freq")
  
  #plot correlations visually
  # corrplot(mtx_corr, type="upper", is.corr=FALSE, tl.col="black", na.label=" ")
  
  # file <- "LiDAR_vs_LAI2200"
  # file <- "LiDAR_Acquisitions"
  png(paste("D:/Mes Donnees/PhD/Figures/lidar/Correlations/Correlations/Intensity_1m/Correlation_plot_",file,"_sup0_5.png", sep=""),
      width = 1000, height = 743, pointsize=20)
  
  corrplot(mtx_corr, method="color", col=brewer.pal(n=8, name="PuOr"),  
           type="upper",
           # order="hclust",
           addCoef.col = "white", # Add coefficient of correlation
           tl.col="black", tl.cex = 0.8, #Text label color and rotation
           number.cex = 0.55, # values
           # hide correlation coefficient on the principal diagonal
           diag=FALSE,
           na.label=" ") 
  
  dev.off()
}

onlySignCorr(DF_cor_LiDARvsLAI2200, "LiDAR_vs_LAI2200")

un autre type :

pairs.panels(iris, 
             method = "spearman", #coef de correlation de Spearman à droite
             hist.col = "#00AFBB",
             density = TRUE,  # affiche la courbe de densite
             ellipses = T # show correlation ellipses (interprétable que pour les var quanti)
)

d’autres fcts de corplot : ggcorrplot()

26.1 Evaluer la performance d’une prédiction

26.1.1 Validation croisée (rééchantillonage)

= séparer données en jeux d’entraînement et de test : caret::createDataPartition()

26.2 Processus gaussiens

généraliser la procédure à une quantité infinie de dimensions kernlab::gausspr() -> écart-type des prédictions, donnant une appréciation de la précision du modèle

26.3 Indices de distances et de dissimilarité/Classification ascendante hierarchique CAH/Creation de dendrogrammes/Kmeans (voir TP2 d’ecologie numérique)

26.4 Permanova : anova multivariée