I’m trying to make a box plot with the following arrays
y = array([[16.2, 16.2, 16.2, ..., 15.9, 16.5, 15.5],
...,
[13.8, 14.8, 12.9, ..., 13.3, 16.1, 13.3]])
> y.shape = (110, 40)
x = array(['2023-08-03T08:00:00.000000000', '2023-08-03T09:00:00.000000000',
...
'2023-08-07T20:00:00.000000000', '2023-08-07T21:00:00.000000000'],
dtype='datetime64[ns]')
> x.shape = (110,)
When doing the boxplot
go.Figure(go.Box(
y=y,
x=x,
name=col,
))
The y axis is interpreted as categorical
I’ve checked that y has the correct numerical type and x is a datetime.
What am I doing wrong?
Hey, thanks for the reply.
I’m trying to structure a MRE but I haven’t had the time yet.
I think the problem stems from the fact that plotly does not interpret correctly the array N x M that I’m passing as y. I was under the impression that the default behaviour would be as in matplotlib: you pass a N array as x, an N x M array as y, and M boxplots will be constructed at N locations defined by x.
This is what I want to achieve (in matplotlib, forget about the grey line)
Is there any way to reproduce this in plotly with the arrays in the shape I have?
I had to trust ChatGPT for this one.
Suppose you have dataframe made like this
| time |
temperature_2m |
temperature_2m_member01 |
temperature_2m_member02 |
temperature_2m_member03 |
… |
| 2023-11-13 14:00:00+05:00 |
14.8 |
15.1 |
14.9 |
14.8 |
… |
| 2023-11-13 15:00:00+05:00 |
15.4 |
15.8 |
15.5 |
15.4 |
… |
| 2023-11-13 16:00:00+05:00 |
15.8 |
16.0 |
15.9 |
15.7 |
… |
| 2023-11-13 17:00:00+05:00 |
15.6 |
15.9 |
15.8 |
15.6 |
… |
| 2023-11-13 18:00:00+05:00 |
14.9 |
15.2 |
15.0 |
15.0 |
… |
I can manually make a boxplot at every time step doing
fig = go.Figure()
for index, row in df.set_index('time').iterrows():
fig.add_trace(go.Box(
x=[index] * len(row),
y=row,
showlegend=False,
boxpoints=False,
marker_color='gray',
))
The xaxis created in this figure is a normal time axis, so it’s possible to use it for other scatter plots (what I wanted to do eventually).
