here I implement proposed changes, most importantly:
optimize_design() returns a new BatchContainer (a copy)
assign_*() functions now also return copies of the BatchContainer
the score is now passed directly to the optimize_design(), and are not being part of the BatchContainer. assigning bc$scoring_f will raise an error
the trace is now stored as a tibble in bc$trace. every row of this tibble is one optimization; and scores and aggregated_scores are now list-columns in this tibble. no more OptimizationTrace
bc$plot_trace() can be used to plot traces, and bc$score_table() returns a table ready to be used in ggplot
Multi-plate optimization seems to work fine.
I did make sure that the following vignettes are compiling successfully:
NCS22_talk.Rmd
optimizer_examples.Rmd
custom_shuffle.Rmd
basic_examples.Rmd
I didn't update tests yet, I would like to hear your feedback first (new API, function naming & so on).
Could you please take a look and let me know if the new API looks good to you? Maybe you could also check if it works in practice for you. Once the API approved, I can start updating other vignettes, tests and so on.
More specifically, what do you think about:
columns in the bc$trace
bc$plot_trace() and bc$scores_table() syntax
bc$score(scoring) and scoring= in optimize_design()
new "read-only" approach in optimize_design() and assign_*()
updated all the vignettes (including cached) and all the tests pass now. should be ready to merge, unless some problems are discovered.
@ingitwetrust @banfai @julianesiebourg
here I implement proposed changes, most importantly:
optimize_design()
returns a newBatchContainer
(a copy)assign_*()
functions now also return copies of the BatchContaineroptimize_design()
, and are not being part of the BatchContainer. assigningbc$scoring_f
will raise an errorbc$trace
. every row of this tibble is one optimization; and scores and aggregated_scores are now list-columns in this tibble. no moreOptimizationTrace
bc$plot_trace()
can be used to plot traces, andbc$score_table()
returns a table ready to be used in ggplotMulti-plate optimization seems to work fine.
I did make sure that the following vignettes are compiling successfully:
I didn't update tests yet, I would like to hear your feedback first (new API, function naming & so on).
Could you please take a look and let me know if the new API looks good to you? Maybe you could also check if it works in practice for you. Once the API approved, I can start updating other vignettes, tests and so on.
More specifically, what do you think about:
bc$trace
bc$plot_trace()
andbc$scores_table()
syntaxbc$score(scoring)
andscoring=
inoptimize_design()
optimize_design()
andassign_*()
Here's a small example:
closes #17, closes #19