problem-12.8

problem-12.8  This model is actually done in the help page for the data set. We can fit the model by running the command
> library(MASS)                 # load in help page
> example(wtloss)
> wtloss.fm
Nonlinear regression model
  model:  Weight ~ b0 + b1 * 2^(-Days/th)
   data:  wtloss
    b0     b1     th
 81.37 102.68 141.91
 residual sum-of-squares:  39.24
> plot(Weight ~ Days, wtloss)   # make scatterplot
> lines(predict(wtloss.fm) ~ Days, data = wtloss)
    
The value of b0 is the predicted long-term weight. If we want the predicted amount after 365 days we have
> predict(wtloss.fm, newdata=data.frame(Days=365))
[1] 98.64

(Of course, predictions beyond the range of the data are not a good idea.)