Displaying 1 result from an estimated 1 matches for "lowervals".
Did you mean:
lowercall
2011 Dec 11
1
nls start values
...t works for some cases, but not for others.
In the example below, the fit works using phase=pi, but not using
phase=0. But there are many examples which fit just fine using 0.
Is there a comparable alternative to nls that is not so extremely
influenced by the start values?
# Data for example fit
lowervals <- list(phase=0, amp=0)
uppervals <- list(phase=2*pi, amp=2)
afreq <- 1 / (24 / 2 / pi)
gene_expression <- c(1.551383, 1.671742, 1.549499, 1.694480, 1.632436,
1.471568, 1.623381,
1.579361, 1.809394, 1.753223, 1.685918, 1.754968, 1.963069, 1.820690,
1.985159, 2.205064,
2.160308, 2.120189...