```{r} library(data.table) library(reshape2) library(ggplot2) allee <- read.csv("allee_data.csv")[2:3] allee_null <- read.csv("allee_nulldata.csv")[2:3] ou <- read.csv("ou_data.csv")[2:3] ou_null <- read.csv("ou_nulldata.csv")[2:3] df <- melt(list(OU = list(conditional=ou, null=ou_null), Allee = list(conditional=allee, null = allee_null)), id=c("variable", "value")) names(df) = c("variable", "value", "data", "model") write.csv(df, "Figure1.csv") df <- read.csv("Figure1.csv") ``` ``` {r Figure1, dev=c("CairoPDF", "CairoPS", "CairoPNG"), fig.width=6, fig.height=4, include=FALSE} ggplot(df, aes(value, y=..density..)) + geom_histogram(data = subset(df, data=="conditional"), binwidth=0.3, alpha=.5) + geom_density(data = subset(df, data=="null"), adjust=2) + facet_grid(model~variable, scales="free_y") + xlim(c(-1, 1)) + xlab("Kendall's tau") + theme_bw() ```` ```{r} library(grid) library(earlywarning) ``` Some reorganization of the data... ```{r} roc_me <- function(dat, nulldat){ test_variance = subset(dat, variable=="Variance") null_variance = subset(nulldat, variable=="Variance") test_autocorrelation = subset(dat, variable=="Autocorrelation") null_autocorrelation = subset(nulldat, variable=="Autocorrelation") roc_var <- roc_data(NULL, null=null_variance$value, test=test_variance$value) roc_acor <- roc_data(NULL, null=null_autocorrelation$value, test=test_autocorrelation$value) rocs <- melt(list(variance = roc_var, autocorrelation = roc_acor), id = c("Threshold", "True.positives","False.positives")) names(rocs)[4] = "Indicator" rocs } allee_roc <- roc_me(allee, allee_null) ou_roc <- roc_me(ou, ou_null) ``` ```{r} ggplot(allee_roc) + geom_line(aes(False.positives, True.positives, col=Indicator)) + theme_bw() ggplot(ou_roc) + geom_line(aes(False.positives, True.positives, col=Indicator)) + theme_bw() ```