trophia / influ Goto Github PK

The .$indices data.frame should be ordered by integer levels of the focus term. But somewhere these integer levels are getting converted too characters. This affects the order in .$indices and thus in stanPlot. Care needs to be taken to have indices merged into .$indices correctly.

Hi Nokome,
The functions stanPlot' andstepPlot' are not working for me, primarily because the plot' command sets the xlim to integers 1,...,n while thelines' commands sets x to integer values of the level field, which are often years (e.g., 2005,...,2013).
Also, I may have only four years spanning 2005 to 2013, for example, and so I like to see the gaps where years are missing. I've attached a simple adaptation of your stanPlot code that works for me (not using proto), only to illustrate what I need to do to make the function work (on Windows 7).
Thanks Nokome.
P.S. Paul Starr uses influence plots all the time, and so should I.
stanPlot.zip

Currently, CDI plots for continuous variables use points with error bars (by cutting the values into bins). Would be less confusing to use lines.

Thanks for creating this useful package!
But I have some problem about it. I used this package to standardize the CPUE of fishery data.
Does this package work for the GLM which retain interaction terms?
If not, could anybody suggest the resolutions of this problems?

Thank you so much!