| Author |
Comment/Response |
David Belisle
|
 04/21/97 11:17pm
Reply to message #307 from Sean Dempsey: > When I fit a function to
a data set I can't seem to get a value for how much uncertainty is
in the function. For example I need to use the slope of a fitted function
as one piece of data in a larger analysis and then do an error propogation;
I need to know that I can trust my slope +/- some level of uncertainty,
how do I determine this? One good way is to use the LinearRegression
package (<<Statistics`LinearRegression`). This package provides information
about the fit, including the sum of squares total, error and model,
the correlation coefficient (r^2). A good estimate for the errors
could be provided using confidence intervals (include the option RegressionReport->ConfidenceInterval
or something like that, and specifying the confidence level with SignificanceLevel->.05
[standard for 95]). I think that this is the best that you can do
in this situation, as the actual errors in the values should be taken
into account based on your method of data acquisition. These parameters
should help you pick a the best model however. I hope that this is
of some help, Dave (david.belisle@utoronto.ca)
URL: , |
|
