Page MenuHomePhabricator

report text
Open, NormalPublic

Description

There have been many little fixes lately in the report text if I saw correctly.
The report texts are quite outdated and represent the state from September last year and was rather preliminary.
Rather than little fixes, it would be good to revise the report text entirely....

Event Timeline

Let's use this task to collect what exactly needs to be revised.

  1. The cluster analysis only makes sense for more than two tasks. This remark is given in the report along with the descriptions of the plots that are not generated when less than three tasks are present. Do we want to keep it like that or omit the whole section?
aguilera added a comment.EditedJun 9 2020, 1:22 PM
  1. The spacing of the record makes that sometimes plots are shown in the following page, the formatting and the spacing should be revised.
  1. When many algorithms are ranked, the first algorithm is not shown in the legend of the plots for section 2.4. (This happened when running the data from the Decathon, where algorithm "Isensee" was missing in the legend of the plots for section 2.4)
  1. In the plots of section 3.3.1 the legend is cut off

Item 3 and 4 are already mentioned in separate tasks: T27474 and T27475

items 3 and 4 solved (see separate tasks).
solved at least for 19 algorithms.

we could decide to put algorithm names in the legend into 2 columns if there are e.g. more than 18 algorithms, using code below. might be problematic/need more adaptations (cut from the right) in case of long algorithm names. would then make sense to do this in all figures.

scale_color_manual(values=cols,
                      guide = guide_legend(ncol = ceiling( length(unique(dd[[1]]$alg_name)) / 18))
                      )