rolling linear regression in R to find optimal fit

Question

I am working with a bacterial growth. There is a point at which these grow exponentially resulting in a straight line when you take the logarithmic value.

Now I want to use an algorithm to automatically detect when they exit out of exponential growth. I decided to try and write a recursive function that monitors for changes in r-squared value.

library('datasets')
library('tidyverse')
data(iris)
summary(iris)

iris %>% ggplot(aes(x=Sepal.Length, y=Petal.Length))+
  geom_point()+
  geom_smooth(method = lm)


x <- 1
rolling_func <- function(df) {
  rsquare_1 <- summary(lm(Petal.Length[x:x+5]~Sepal.Length[1:x], data = iris))[[8]]
  rsquare_2 <- summary(lm(Petal.Length[x+5:x+10]~Sepal.Length[1:x], data = iris))[[8]]
  if (rsquare_2 < rsquare_1){
    return(x)
  } else {
    x <- x+5
    rolling_func(df)
  }
}

rolling_func(iris)

However when I try to run this I get the following error

Error: C stack usage 7969828 is too close to the limit

Basically what I want my code to do is check when I no longer have a linear correlation between x and y

Thank you very much for your help.

Answer 1

Here's an approach using dplyr and a for loop

library(dplyr)

chunksize <- 3
running <- vector(length = nrow(iris)/chunksize)

for (i in 1:(nrow(iris)/chunksize)) {
   abc <- iris %>% 
      slice(1:(i * chunksize)) %>% 
      lm(Petal.Length ~ Sepal.Length, data = .)
   running[[i]] <- summary(abc)$r.squared
   if (i > 1) {
      if (running[[i]] < running[[i - 1]]) {
         break
      }
   }
   print(paste("NUmber of iris rows after which r squared goes down is",
               i*chunksize))
}

#> [1] "NUmber of iris rows after which r squared goes down is 3"

running

#>  [1] 0.7500000 0.3963221 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#>  [8] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [15] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [22] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [29] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [36] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [43] 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
#> [50] 0.0000000