Multiple stat_function on grouped data with ggplot2

Question

I am studying a data set with multiple observation of a parameter overtime. the data is like:

test<-data.frame(t = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0.33, 0         


        

Answer 1

I thought of another naive solution: Predicting the values of every dlnorm().

## Split up the data according to t
tt     <- split(test, test$t)

## Fit a lognormal to every dataset
fits   <- lapply(tt, function(x) fitdistr(x$int, "lognormal"))

## Predict values
fitted <- lapply(fits, function(x) dlnorm(x = 1:max(test$int),
                               mean = x$estimate[1], sd = x$estimate[2]))

## Wrap everything into a data.frame ggplot can handle
plot.data <- data.frame(y = unlist(fitted), int = 1:max(test$int),
                        t = rep(unique(test$t),
                            each = length(unlist(fitted))/length(unique(test$t))))

## Plot
ggplot(test, aes(int, group = as.factor(t), colour=t)) +
  #geom_density() +
  geom_line(data = plot.data, aes(y = y), lwd = 1)

Answer 2

What about a naive solution, adding iteratively stat_function()?

    cols <- brewer.pal(length(unique(test$t)),"Set1")
    g <- ggplot(data=subset(test, t == 0, select='int'), aes(x=int))
    n <- 1
    for(i in unique(test$t)){ 
        fit <- fitdistr(subset(test, t == i, select='int')$int, "lognormal")
        g <- g+stat_function(fun = dlnorm, 
                             args=list(mean=fit$estimate[1],sd=fit$estimate[2]), 
                             col=cols[n])
        n <- n + 1
    }
    g