[R] Logistic Regression output baseline (reference) category
David Winsemius
dwinsemius at comcast.net
Sat Mar 26 07:39:13 CET 2016
> On Mar 25, 2016, at 10:19 PM, Michael Artz <michaeleartz at gmail.com> wrote:
>
> Hi,
> I have now read an introductory text on regression and I think I do
> understand what the intercept is doing. However, my original question is
> still unanswered. I understand that the intercept term is the constant
> that each other term is measured against. I think baseline is a good word
> for it. However, it does not represent any one of the x variables by
> itself.
It represents all of the X variables at their reference levels. There are no individual intercepts on a variable-by-variable basis. You accepted the notion of "baseline" You cannot parcel out single variable intercepts. What would they actually mean anyhow?
> Is there a way in R, to extrapolate the individual x variable
> intercepts from the equation somehow.
If you describe the process by which that could be calculated, we might have basis for discussion, but as it is I think you still need to be studying the theory more. I don't intend any further resspones on R-help where these questions are off-topic, so you should direct any further questions to stats.stackexchange.com
>
>
> On Tue, Mar 15, 2016 at 8:26 PM, David Winsemius <dwinsemius at comcast.net>
> wrote:
>
>>
>>> On Mar 15, 2016, at 1:27 PM, Michael Artz <michaeleartz at gmail.com>
>> wrote:
>>>
>>> Hi,
>>> I am trying to use the summary from the glm function as a data source.
>> I
>>> am using the call sink(<some file>) then
>>> summary(logisticRegModel)$coefficients then sink().
>>
>> Since it's a matrix you may need to locate a function that write matrices
>> to files. I seem to remember that the MASS package has one.
>>
>>> The independent
>>> variables are categorical and thus there is always a baseline value for
>>> every category that is omitted from the glm output.
>>
>> Well, it's not really omitted, so much as shared among all variables. For
>> further reading in the halp pages consult:
>>
>> ?model.matrix
>> ?contrasts
>> ?contr.treatment
>>
>> But you probably need to supplement that with an introductory text that
>> covers R regression defaults.
>>
>>> I am interested in how
>>> to get the Z column for all of the categorical values.
>>
>> The Z column? You meant the "z value" column. Again, since it's a matrix
>> you need to use column indexing with "["
>>
>> summary(logisticRegModel)$coefficients[ , "z value"]
>>
>> Read up on the summary function for glm objects at:
>>
>> ?summary.glm
>>
>>
>>> I don't see any row
>>> for the reference category.
>>
>> What do you imagine the (Intercept) row to be doing? If you are having
>> difficulty understanding this (which is not really an R-specific issue)
>> there are probably already several explanations to similar questions on:
>>
>> http://stats.stackexchange.com/
>>
>>
>>>
>>> How can I get this Z value in the output?
>>
>> Asked and answered.
>>
David Winsemius
Alameda, CA, USA
More information about the R-help
mailing list