Base model being no reason given
Could you please elaborate on how the base reference is when there is no reason (aka reason 0)? I'm confused about the interpretability of the odds ratio of the reasons. Why do we use absent people with reason 0 as the base reference?
For example, if we took another feature e.g. children, wouldnt the interpretation be: for each child the absent person has, they're 1.5 times more likely to be excessively absent. Or would it be the absent person with children is 1.5 times more likely to be excessively absent compared to those absent with no children? Intuitively, the former makes more sense to me as it indicates the linearity.
Hi Kai N!
Thanks for reaching out.
This is the moment to apply some understanding and economic reasoning when taking our decisions. Please refer to the following text lecture from the course for a detailed explanation: https://learn.365datascience.com/courses/sql-tableau-python/dropping-a-dummy-variable/
That's why we don't choose the children example.
If it were to chose "children" - no, this reasoning will not apply for a discrete numeric variable! Here we are talking about dummy variables, so the dummy variable there (which is not the baseline interpretation) would be "a person having one or more children, is more likely to... by ...".
Hope this helps.
Kind regards,
Martin
