0

I'm running a for loop to do bulk nonlinear regression. The problem is, some columns end up causing an error because there aren't enough iterations. 

Error in nls(y ~ cbind(1, 1/(1 + exp((xmid - x)/exp(lscal)))), data = xy,  : 
  number of iterations exceeded maximum of 50

I tried changing the number of iterations (the following) but it still gives me the same error. I'm not sure what to do. Either I need it to flag the columns that don't run or to iterate until it converges. The strange thing is that for all the ones that run, they take only 1 iteration so I don't understand how this can go over 50? 

for(i in 1:(ncol(df))) {jpeg(paste((i), ".jpg"), height = 5, width = 6, units = "in", res = 300)
    plot(time, 
         df[[i]],
         xlab = "time (hours)",
         ylab = "fluorescence (rfu)",
         xlim=c(0, 92), ylim=c(0, 150000),
         pch = 20,
         col = "grey65",
         main=colnames(df[i]))
    N$i <- nlsLM(df[[i]]~SSfpl(time, a, b, c, d), control=nls.lm.control(maxiter=100, nprint = 1))
    lines(time, predict(N$i), col = "black")
    dev.off()
}

Any help would be greatly appreciated!

79edinburgh
  • 1
  • 1

1 Answers1

0

I did the control= part differently. Try this: Specifically, I have :

maxiter <- 32000
control = list(maxiter = maxiter, gtol=converge)
edwardmoradian
  • 53
  • 10