Plotly boxplot: groupby option?

Question

I have two boolean variables and I try to create a boxplot with groups. Each group should represent one of the variables and it should contain two boxplots, one for TRUE and one for FALSE. Instead I am getting two groups, one representing TRUE and one representing FALSE and for each group two boxplots corresponding to each variable as in the attached graph:

I understand that groups are derived from the xaxis. But how can I make plotly think that the variable names are the groups? The code I used for the output :

trace3= Box(
 y=raw_matrix.TPS,
 x=raw_matrix.noClassGc,
 name='noClassGc',
 marker=Marker(
 color='#3F5D7D'
))

trace4= Box(
 y=raw_matrix.TPS,
 x=raw_matrix.aggresiveOpts,
 name='aggresiveOpts',
 marker=Marker(
 color='#0099FF'
 ))

data = Data([trace3, trace4])
layout = Layout(
 yaxis=YAxis(
 title='confidence',
 zeroline=False),
 boxmode='group',
 boxgroupgap=0.5
 )


fig = Figure(data=data, layout=layout)
plot_url = ploteczki.plot(fig, filename='Performance by categoricals parameters')

Answer 1

You need to rearrange your data arrays so that the two Box traces have 'x' coordinates of 'noClassGc' and 'aggresiveOpts'.

This IPython notebook shows you how to do so.

Answer 2

Alternatively, to represent each group by its boolean boxplots, you can assign each trace to different x-axes. Here is an example:

trace0 = Box(
    y=raw_matrix_TPS,
    x=raw_matrix_noClassGc,
    name='noClassGc',
    marker=Marker(
        color='#3F5D7D'
    )
)
trace1 = Box(
    y=raw_matrix_TPS,
    x=raw_matrix_aggresiveOpts,
    name='aggresiveOpts',
    xaxis='x2',
    marker=Marker(
        color='#0099FF'
    )
)
data = Data([trace0, trace1])
layout = Layout(
    xaxis = XAxis(
        domain=[0, 0.55],
    ),
    xaxis2 = XAxis(
         domain=[0.55, 1],
    ),
    yaxis = YAxis(
         title='confidence',
         zeroline=False
    ),
    boxmode='group',
    boxgroupgap=0.5
)
fig = Figure(data=data, layout=layout)
plot_url = py.plot(fig, filename='Performance by categoricals parameters')

Here is the link to the plot

To learn more, you can checkout Plotly Python reference