feat(cross-validation): Split time-related results into their own plots #986
Conversation
augustebaum
force-pushed
the
902-split-plots
branch
from
0fe1e91 to
6b0603e
Compare
| def linspace(lo, hi, num): | ||
| interval = (hi - lo) / (num - 1) | ||
| return [lo + k * interval for k in range(0, num)] |
There was a problem hiding this comment.
Maybe you can use numpy.linspace here ?
There was a problem hiding this comment.
I did originally, but re-implementing it avoids bringing in numpy as a direct dependency for just 1 function
There was a problem hiding this comment.
...although I see that we import numpy in cross_validation_item.py...
There was a problem hiding this comment.
And numpy is a direct dependency of sklearn which is our direct dependency. ♻️
skore/src/skore/sklearn/cross_validation/plots/compare_scores_plot.py
Outdated
Show resolved
Hide resolved
skore/src/skore/sklearn/cross_validation/plots/timing_normalized_plot.py
Outdated
Show resolved
Hide resolved
| def linspace(lo, hi, num): | ||
| interval = (hi - lo) / (num - 1) | ||
| return [lo + k * interval for k in range(0, num)] |
There was a problem hiding this comment.
Same as above could you use numpy.linspace ?
| # FIXME: Maybe this logic belongs in CrossValidationPlots | ||
| plots_bytes = { | ||
| plot_name: plotly.io.to_json(plot, engine="json").encode("utf-8") | ||
| for plot_name, plot in dataclasses.asdict(plots).items() |
There was a problem hiding this comment.
Could the plot name be "humanized" ?
timing_normalized => Normalized timings ?
There was a problem hiding this comment.
You mean just in the front-end right? I mean, in their Python code user would still have to write
reporter.plots.timing_normalizedThere was a problem hiding this comment.
How should the other plots be renamed? Maybe something like
- compare_scores -> "Scores"
- timing -> "Timings"
- normalized_timing -> "Normalized timings"
@MarieS-WiMLDS Does this make sense?
There was a problem hiding this comment.
it completely makes sense! I like that it's easy from the human name to find how to call it in python. With that logic, maybe we could even switch from compare_scores to just scores?
Closes #902
Screen-record:
2024-12-18T15_00_29_screen_record.mp4
plotsattributeplot_cross_validationfunction