using a custom function for smoothing

Question

I recently started learning R and ran into a problem using a custom function when plotting. I want to use the rosin-rammler equation but can't figure out how to use it in my case. In Mathkad, this was easily solved. And here I can not understand.

This is an example of my data and code.

xy	yy	cat1	sem1
0	0	0	№1
10	3	0	№1
20	30	0	№1
30	18	0	№1
40	15	0	№1
50	13	0	№1
60	8	0	№1
70	5	0	№1
80	3	0	№1
90	2	0	№1
100	0	0	№1
110	0	0	№1

ggplot(allOnw, aes(x = xy, y=yy, col = sem1)) +geom_point()+
  guides(fill = FALSE) +  labs(col="S" )+theme(legend.position="bottom")  +scale_x_continuous(breaks = seq(from = 0, to = 160, by = 10), limits = c(0,160))+
  scale_color_manual(values = c("blue", "cyan", "darkgreen","darkorange","gold","darkorchid")) + 
  facet_wrap(vars(cat1),ncol = 1, strip.position = "right")+labs(x ="MP")+labs(y = "GT ")

I would like the graph to end up looking like when using geom_density. What can i do for this? Is it possible?

Answer 1

If you have a particular model to fit to your points, you can use method = nls inside geom_smooth.

As far as I can tell from the literature, the function you are trying to fit takes this form:

rosin_rammler <- function(x, A, B, C) {
  C * ((A / x) * (x / B)^A * exp(-((x / B)^(A - 1))))
}

And we can use this directly inside geom_smooth:

ggplot(allOnw, aes(x = xy + 1e-6, y = yy, col = sem1)) +
  geom_point() +
  geom_smooth(formula = y ~ rosin_rammler(x, A, B, C), method = nls,
              method.args = list(algorithm = "port",
                                 start = list(A = 5, B = 20, C = 1000), 
                                 lower = c(1, 1)), se = FALSE, n = 1000) +
  labs(col = "S", x = "MP", y = "GT")+
  theme(legend.position = "bottom")  +
  scale_x_continuous(breaks = 0:16 * 10, limits = c(0, 160)) +
  scale_color_manual(values = c("blue", "cyan", "darkgreen", 
                                      "darkorange", "gold", "darkorchid")) + 
  facet_wrap(vars(cat1), ncol = 1, strip.position = "right")