I have a dataset named daph
daph <- read.table(text = "t v
20 19
25 78.2
30 254.8
",header = TRUE, sep = "")
and what I'm trying to do is adding an exponential trend line to a barplot with these values
ggplot(data=daph, aes(x=t, y=v, width=1)) +
geom_bar(stat="identity", fill="steelblue") +
theme_minimal(base_size=18) +
geom_smooth(method = 'nls', formula = y ~ a * exp(b * x), se = FALSE, method.args = list(start = list(a = 1, b = 1)))
but it gives me an error message every time (singular gradient, roughly translated).
I suppose it's got something to do with my starting values, but I'm not enough into maths to understand a lot about this.
Maybe some of you can help me :)
nls are notoriously hard to fit. For your scenario, try providing better starting values or consider using the linear model instead, you can check it out in this post:
lmfit = lm(log(v) ~ t,data=daph)
A = coef(lmfit)[1]
B = coef(lmfit)[2]
ggplot(data=daph, aes(x=t, y=v, width=1)) +
geom_bar(stat="identity", fill="steelblue") +
theme_minimal(base_size=18) +
geom_smooth(method = 'nls', formula = y ~ a * exp(b * x),
se = FALSE, method.args = list(start = list(a = A, b = B)))
The data:
structure(list(t = c(9, 21, 29), v = c(19, 78.2, 254.8)), row.names = c(NA,
-3L), class = "data.frame")