proposal to swap boxplot with density plot

[edarin]

Proposal concerning the visualisation of covariates mismatch

To me, the boxplots shown are focusing our attention to the outlier observations.
Thus neither covariates seems to be correct

A substitute is proposed in this commit with density plot.

Code is already integrated:

top-down-tutorial/4_tutorial.Rmd

Line 471 in 5bb5b25

# two panel layout

[doug-leasure]

I didn't include the density plots because the code for the plot is a bit more complicated and also the plots don't come out looking good with the default settings for the density function. The plot on the right needs a custom smoothing parameter, and I thought that discussion was beyond the scope of this tutorial.

I thought the boxpots showed the pattern clearly with simple code.