Prediction with nls, always return prediction values of train data set

Question

I have 4 points of the train set, I use "nls" to fit the train set, then predict the response of test set which contains 2 points. However, the "predict" command always returns values of the train set.

The code is attached below:

##Following is train set
HD = c(714,715,716.6,717.6)
p_l = c(0.5,0.1,0.05826374, 0.005982334)
##Fitting with nls
raw_data = data.frame(HD,p_l)
exp_fit = nls(p_l~exp(a+b*HD),data = raw_data,trace = T,start = list(a = 0,b 
= 0))
##Following is test set
HD_test = c(718.2,719.17)
p_l_test = predict(exp_fit,newdata = HD_test)

Answer 1

You need to pass either a data frame or a named list as newdata.

Use the same name as your original predictor (HD). Otherwise, newdata is treated as missing, and fitted values from the original training data are returned.

From the nls docs:

newdata
A named list or data frame in which to look for variables with which to predict. If newdata is missing the fitted values at the original data points are returned.

HD_test = data.frame(HD = c(718.2,719.17)) # wrap in data frame, name "HD"
p_l_test = predict(exp_fit, newdata = HD_test)

p_l_test
[1] 0.0009320202 0.0002184518