Best function for modelling diminishing returns - r

Question

I am visiting a bird sanctuary that has many different species of birds. Some species are more numerous while other species are less numerous. I came back to the sanctuary 9 times and after every visit I am calculating the total number of species I observed. Unsurprisingly, there is a diminishing return in my visits, since I observe the most numerous species on my every visit, but it does not increase the count of observed species. What is the best function in R to predict how many birds I will observe on my 20th visit?
Here is the data.frame
d <- structure(list(visit = 1:9,
totalNumSpeciesObserved = c(200.903, 296.329, 370.018, 431.59, 485.14, 533.233, 576.595, 616.536, 654)),
class = "data.frame", row.names = c(NA, 9L))
I expect to see a model that fits data well and behaves in a "log-like" fashion, predicting diminishing returns

Answer 1

In order to best ask a question, stack has some good links: https://stackoverflow.com/help/how-to-ask
If you're trying to model this, I might take the approach of a regression on the square root of the independent variable based on the data. Kind of strange to think about it as a function of visits though... Maybe if it were even spaced time periods it would make more sense.
d <- structure(list(visit = 1:9,
totalNumSpeciesObserved = c(200.903, 296.329, 370.018, 431.59, 485.14, 533.233, 576.595, 616.536, 654)),
class = "data.frame", row.names = c(NA, 9L))
mod <- lm(totalNumSpeciesObserved ~ I(sqrt(visit)), d)
new.df <- data.frame(visit=1:13)
out <- predict(mod, newdata = new.df)
plot(d, type = 'o',pch = 16, xlim = c(1,13), ylim = c(200,800), lwd = 2, cex = 2)
points(out, type= 'o', pch = 21, col = "blue", cex = 2)
The I() wrapper allows you to transform the independent variable on the fly, hense the use of sqrt() without needing to save a new variable.

Answer 2

I also don't know if this helps, but you could build a simulator to test for asymptoptic behaviour. For example you could build a population:
population <- sample(size = 1e6, LETTERS[1:20],
replace = TRUE, prob = 1/(2:21)^2)
This would say there are 20 species and decreasing probability in your population (expand as you wish).
The you could simulate visits and information about your visit. For example how large is the sample of your visit? During a visit you only see 1% of the rainforest etc.
sim_visits <- function(visits, percent_obs, population){
species_viewed <- vector()
unique_views <- vector()
for(i in 1:visits){
my_samp <- sample(x = population, size = round(percent_obs*length(population),0),
replace = FALSE)
species_viewed <- c(species_viewed, my_samp)
unique_views[i] <- length(unique(species_viewed))
}
new_observed <- unique_views - dplyr::lag(unique_views, 1, 0)
df <- data.frame(unique_views = unique_views, new_observed)
df$cummulative <- cumsum(unique_views)
df
}
And then you could draw from the simulation many times and see what distribution of values you get.
sim_visits(9, percent_obs = .001, population = population)
unique_views new_observed cummulative
1 13 13 13
2 15 2 28
3 15 0 43
4 17 2 60
5 17 0 77
6 17 0 94
7 17 0 111
8 17 0 128
9 17 0 145
And don't know if this is helpful, but I find simulation a good way to conceptualise problems like these.