neleval plot-systems
¶
Summarise system results as scatter plots
Usage summary¶
$ neleval plot-systems --help
usage: neleval plot-systems [-h] [--by-system | --by-measure | --single-plot]
[--scatter | --rows | --columns | --heatmap]
[--pr | --prf | --recall-only] [--lines]
[--cmap CMAP] [--limits LIMITS]
[-i {evaluate,confidence}]
[-o OUT_FMT | --interactive [SHELL] | --run-code
CODE] [--figsize FIGSIZE]
[--legend-ncol LEGEND_NCOL] [-m NAME]
[--ci CONFIDENCE] [--group-re GROUP_RE]
[--best-in-group [BEST_IN_GROUP]] [-s SORT_BY]
[--at-most AT_MOST] [--label-map LABEL_MAP]
[--style-map STYLE_MAP] [--anon]
FILE [FILE ...]
Summarise system results as scatter plots
positional arguments:
FILE
optional arguments:
-h, --help show this help message and exit
--by-system Each system in its own figure, or row with --heatmap
--by-measure Each measure in its own figure, or row with --heatmap
(default)
--single-plot Single figure showing fscore for all given measures
--scatter Plot precision and recall as separate axes with
different markers as needed
--rows Show rows of P/R/F plots
--columns Show columns of P/R/F plots (default)
--heatmap Show a heatmap comparing all systems and measures
--pr In rows or columns mode, plot both precision and
recall, rather than F1
--prf In rows or columns mode, plot precision and recall as
well as F1
--recall-only
--lines Draw lines between points in rows/cols mode
--cmap CMAP
--limits LIMITS Limits the shown score range to the specified min,max;
or "tight"
-i {evaluate,confidence}, --input-type {evaluate,confidence}
Whether input was produced by the evaluate (default)
or confidence command
-o OUT_FMT, --out-fmt OUT_FMT
Path template for saving plots with --fmt=plot
(default: ./{}.pdf))
--interactive [SHELL]
Open an interactive shell with `figures` available
instead of saving images to file
--run-code CODE Run the given Python code with `figures` available
instead of saving images to file
--figsize FIGSIZE The width,height of a figure in inches (default 8,6)
--legend-ncol LEGEND_NCOL
Number of columns in legend; otherwise ensures at most
20
-m NAME, --measure NAME
Which measures to use: specify a name (or group name)
from the list-measures command. This flag may be
repeated.
--ci CONFIDENCE The percentile confidence interval to display as error
bars (requires --input-type=confidence
--group-re GROUP_RE Display systems grouped, where a system's group label
is extracted from its path by this PCRE
--best-in-group [BEST_IN_GROUP]
Only show best system per group, optionally according
to a given measure
-s SORT_BY, --sort-by SORT_BY
Sort each plot, options include "none", "name",
"score", or the name of a measure.
--at-most AT_MOST Show the first AT_MOST sorted entries
--label-map LABEL_MAP
JSON (or file) mapping internal labels to display
labels
--style-map STYLE_MAP
JSON (or file) mapping labels to <color>/<marker>
settings
--anon Hide system/team names