I ran into this issue while running regressions. The coefficient values in the output of the tool only go out to six digits, but depending on the units used on my input variables, sometimes the values of the output coefficients can be so small that they show up as zero. However, even when my calculated coefficients are really small, they tend to increase once I standardize them (this may not always apply but it does with the data I'm currently working on. The units of my inputs do not affect the final standardized coefficient, but they do affect the unstandardized coefficient, and when the unstandardized coefficient is very small (or even zero), some precision is lost when calculating my standardized coefficient. Presenting the output results of the Generalized Linear Regression tool, as well as other spatial analyst/statistics tools, as an equal number of significant figures rather than decimal places would help solve this issue.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.