如何在Plotly堆叠柱状图中设置单侧误差棒？

我使用Plotly绘制了堆叠柱状图，当前图表的误差棒为双向样式，希望将其修改为单侧误差棒（效果参考经Illustrator修改后的图表）。

当前实现代码 #

import plotly.graph_objects as go

y1_values = [5, 4, 4, 5, 6, 5]
y2_values = [9, 8, 7, 6, 5, 4]
y3_values = [4, 6, 7, 7, 7, 9]

y1_errors = [0.5, 0.6, 0.4, 0.7, 0.5, 0.8]
y2_errors = [0.4, 0.3, 0.5, 0.7, 0.3, 0.5]
y3_errors = [0.6, 0.2, 0.3, 0.4, 0.4, 0.4]

fig = go.Figure()

# Bars for increase with error bars
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y3_values, name="Increase", marker_color='#FF6692',
            error_y=dict(type="data", array=y3_errors, thickness=2, width=10),
            base=[sum(x) for x in zip(y1_values, y2_values)]
            )

# Bars for decrease with error bars
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y2_values, name="Decrease", marker_color='#19D3F3',
            error_y=dict(type="data", array=y2_errors, thickness=2, width=10),
            base=y1_values
            )

# Bars for unchanged with error bars
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y1_values, name="Unchanged", marker_color='#00CC96',
            error_y=dict(type="data", array=y1_errors, thickness=2, width=10),
            base=[0] * 6
            )

# Annotation for bar1 and bar2
fig.add_annotation(text="10 minutes",
                   xref="paper", yref="paper",
                   x=0.07, y=-0.18,
                   showarrow=False,
                   font_size=20
                   )

# Annotation for bar3 and bar4
fig.add_annotation(text="30 minutes",
                   xref="paper", yref="paper",
                   x=0.5, y=-0.18,
                   showarrow=False,
                   font_size=20
                   )

# Annotation for bar5 and bar6
fig.add_annotation(text="60 minutes",
                   xref="paper", yref="paper",
                   x=0.93, y=-0.18,
                   showarrow=False,
                   font_size=20
                   )

# Layout
fig.update_layout(barmode="stack", showlegend=True, template='plotly_white',
                  height=700,
                  width=1000,
                  legend=dict(title='', orientation="h", traceorder='normal', x=0.46, y=1.05,
                              bgcolor='rgba(0,0,0,0)', bordercolor='rgba(0,0,0,1)', borderwidth=0,
                              font_size=20
                              )
                  )

fig.update_yaxes(showline=True, showgrid=False, linewidth=0.5, linecolor='black',
                 title='<b>Percent</b>', titlefont=dict(size=24), title_standoff=30,
                 ticks='outside',
                 dtick=2,
                 ticklen=10,
                 tickfont=dict(size=20),
                 range=[0, 20]
                 )

fig.update_xaxes(title='', tickvals=[1, 1.5, 2.2, 2.7, 3.4, 3.9],
                 ticktext=["<b>Cat</b><br>n = 15", "<b>Dog</b><br>n = 15",
                           "<b>Cat</b><br>n = 15", "<b>Dog</b><br>n = 15",
                           "<b>Cat</b><br>n = 15", "<b>Dog</b><br>n = 15"],
                 ticks="", tickfont_size=20, linecolor="black", linewidth=1
                 )

fig.show()

当前双向误差棒效果 #

期望单侧误差棒效果 #

解决方案 #

要实现单侧误差棒，只需在每个error_y参数中添加symmetric=False，并将arrayminus设为全0数组，这样误差棒就只会向上延伸，不会向下。修改后的核心代码如下：

# 修改后的Increase柱子
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y3_values, name="Increase", marker_color='#FF6692',
            error_y=dict(type="data", array=y3_errors, arrayminus=[0]*6, symmetric=False, thickness=2, width=10),
            base=[sum(x) for x in zip(y1_values, y2_values)]
            )

# 修改后的Decrease柱子
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y2_values, name="Decrease", marker_color='#19D3F3',
            error_y=dict(type="data", array=y2_errors, arrayminus=[0]*6, symmetric=False, thickness=2, width=10),
            base=y1_values
            )

# 修改后的Unchanged柱子
fig.add_bar(x=[1, 1.5, 2.2, 2.7, 3.4, 3.9], y=y1_values, name="Unchanged", marker_color='#00CC96',
            error_y=dict(type="data", array=y1_errors, arrayminus=[0]*6, symmetric=False, thickness=2, width=10),
            base=[0] * 6
            )

完整代码只需替换上述三个add_bar部分即可，其他代码保持不变，生成的图表将与期望的单侧误差棒效果一致。

内容的提问来源于stack exchange，提问作者hans