问题
I am new to the function GAMLSS in r, and when I run my code I always get this error: Response Variable out of range
After looking into the data frame, I realized the issue was one of response variables was 0.0000.
I was wondering if someone could explain to me why 0 is out of range and possible solutions to go around it (ex. such as replacement the values)?
回答1:
LOGNO family corresponds to the log-normal distribution, which is defined for positive values only.
The possible solutions might be (but highly depend on the context):
- use another distribution, which better models the response variable and allows zero values
- sometimes zero values are reported if they are below the limit of detection (LOD). In this case, one has a censored data set, and you may look for review of the methods, how to tackle it. A pragmatic approach is to substitute zeros with values like LOD/2, reviewed, for example, here. However, it may result in a very biased estimation.
来源：https://stackoverflow.com/questions/56972789/response-variable-out-of-range-using-gamlss-in-r-logno