an exponential regression is in simply a linear regression between the
xlist and ln(ylist). As ln(0) means trouble... you get the domain
error.
In fact, regressions are "all" linear (except the sinusoidal and
logistic): quadratic, cubic and quartic correspond to several (2, 3 or
4 resp.) linear regressions with dummy variables, while exponential,
power and log regressions are linear regressions with some
manipulation of the original data lists ( xlist vs. ln(ylist), ln(x)
vs. ln(y) and xlist vs. exp(ylist) resp.)
Cheers,
Nelson