ValueError('y must be real') when plotting by + rasterize on a sparse dataset #1352
Description
Not entirely sure if this is datashader or hvplot so feel free to transfer over, but I'd expect this to work.
import hvplot.pandas
from bokeh.sampledata.penguins import data as df
df.hvplot.points(
x="bill_length_mm",
y="bill_depth_mm",
by="species",
rasterize=True
)File ~/miniconda3/envs/census/lib/python3.12/site-packages/holoviews/operation/datashader.py:422, in aggregate._apply_datashader(self, dfdata, cvs_fn, agg_fn, agg_kwargs, x, y)
417 with warnings.catch_warnings():
418 warnings.filterwarnings(
419 action='ignore', message='casting datetime64',
420 category=FutureWarning
421 )
--> 422 agg = cvs_fn(dfdata, x.name, y.name, agg_fn, **agg_kwargs)
424 is_where_index = ds15 and isinstance(agg_fn, ds.where) and isinstance(agg_fn.column, rd.SpecialColumn)
425 is_summary_index = isinstance(agg_fn, ds.summary) and "index" in agg
File ~/miniconda3/envs/census/lib/python3.12/site-packages/datashader/core.py:474, in Canvas.line(self, source, x, y, agg, axis, geometry, line_width, antialias)
466 if not isinstance(non_cat_agg, (
467 rd.any, rd.count, rd.max, rd.min, rd.sum, rd.summary, rd._sum_zero,
468 rd._first_or_last, rd.mean, rd.max_n, rd.min_n, rd._first_n_or_last_n,
469 rd._max_or_min_row_index, rd._max_n_or_min_n_row_index, rd.where,
470 )):
471 raise NotImplementedError(
472 f"{type(non_cat_agg)} reduction not implemented for antialiased lines")
--> 474 return bypixel(source, self, glyph, agg, antialias=glyph.antialiased)
File ~/miniconda3/envs/census/lib/python3.12/site-packages/datashader/core.py:1333, in bypixel(source, canvas, glyph, agg, antialias)
1330 source, dshape = _bypixel_sanitise(source, glyph, agg)
1332 schema = dshape.measure
-> 1333 glyph.validate(schema)
1334 agg.validate(schema)
1335 canvas.validate()
File ~/miniconda3/envs/census/lib/python3.12/site-packages/datashader/glyphs/points.py:132, in _PointLike.validate(self, in_dshape)
130 raise ValueError('x must be real')
131 elif not isreal(in_dshape.measure[str(self.y)]):
--> 132 raise ValueError('y must be real')
ValueError: y must be real