search

SixiangHu / DataMan

R package for data cleaning, preliminary data analysis and modeling assessing with visualisation.
3 stars 0 forks source link

change of the base of a variable in a glm model may affect the modelPlot #32

Closed SixiangHu closed 8 years ago

SixiangHu commented 8 years ago

Current modelPlot ignores the actual base of variables in a glm, instead, it recalcualtes the mean or mode of the variable and treat them as the base.

If modeller specified another base and used in glm, the modePlot may not recognise it, hence may give wrong "model prediction at base level" line.

SixiangHu commented 8 years ago

The solution could be simple as to let users to specify the base level by passing a list in which includes a name column and base level column.