Error in visualize splines logistic regression with ns in R

Asked Feb 21 '23 at 00:03

Active Feb 21 '23 at 01:31

Viewed 43 times

I have this error when I want to visualize splines logistic regression with ns:

data <- read.table("e:/data_logistic.txt",　header = TRUE)

dd <- datadist(data)

options(datadist = "dd")

model <- lrm(formula = P_A ~ ns(HEIGHT, 4), data = data, x = TRUE, y = TRUE)

p1 <- ggplot(Predict(model, HEIGHT))

Error in quantile.default(x[!outside], knots) : missing values and NaN's not allowed if 'na.rm' is FALSE

When I use rcs (from rsm package) instead of ns (from spines package) I have not any problem. I want to use ns method. How can I solve this problem?

Thanks.

edited Feb 21 '23 at 01:31

asked Feb 21 '23 at 00:03

Mostafa Ahmadi

1

Can we have a [mcve] please? How many unique values of `HEIGHT` do you have? – Ben Bolker Feb 21 '23 at 00:15
The data has 96 observations, 32 presence points and 64 absence points. almost 98% are unique values. – Mostafa Ahmadi Feb 21 '23 at 00:19
1

OK, then my only guess doesn't work. We'll definitely need the example data. – Ben Bolker Feb 21 '23 at 00:20
Could you explain more about example data? – Mostafa Ahmadi Feb 21 '23 at 00:28
1

https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example – Ben Bolker Feb 21 '23 at 00:28
1

even `summary(data$HEIGHT)` would probably help. – Ben Bolker Feb 21 '23 at 00:29
1

I can at least replicate the problem with `set.seed(101); data <- data.frame(HEIGHT = c(runif(95, 150, 650), 1500), P_A = sample(0:1, size = 96, replace = TRUE))` – Ben Bolker Feb 21 '23 at 01:00
Thank you so much but I still have this error. – Mostafa Ahmadi Feb 21 '23 at 01:13
1

This is clearly using the "rms" package. Spelling is crucially important. – IRTFM Feb 21 '23 at 02:57
2

The error is coming from `rms::Predict`. I guess the message is use `rms::rcs` rather than `splines::ns`. Was there any reason to assume that `rms::Predict` should work with `ns`? The rms functions are build to work together but not necessarily work with other packages. – IRTFM Feb 21 '23 at 03:05
You are right about it. I understand now – Mostafa Ahmadi Feb 21 '23 at 09:46

Error in visualize splines logistic regression with ns in R

0 Answers0