Plotly：在散点图分类轴上躲避重叠点
Posted
技术标签:
【中文标题】Plotly：在散点图分类轴上躲避重叠点【英文标题】：Plotly: Dodge overlapping points on scatterplot categorical axis 【发布时间】：2020-11-06 04:30:49 【问题描述】：我正在尝试使用 plotly 来比较回归模型的系数，使用误差条作为置信区间。我使用下面的代码来绘制它，使用变量作为散点图中的分类y 轴。问题是这些点是重叠的，我想像设置barmode='group' 时在条形图中发生的那样避开它们。如果我有一个数字轴，我可以手动躲避它们，但我做不到。
fig = px.scatter(
df, y='index', x='coef', text='label', color='model',
error_x_minus='lerr', error_x='uerr',
hover_data=['coef', 'pvalue', 'lower', 'upper']
)
fig.update_traces(textposition='top center')
fig.update_yaxes(autorange="reversed")
使用构面，我几乎得到了我想要的结果，但有些标签偏离了情节并且不可见：
fig = px.scatter(
df, y='model', x='coef', text='label', color='model',
facet_row='index',
error_x_minus='lerr', error_x='uerr',
hover_data=['coef', 'pvalue', 'lower', 'upper']
)
fig.update_traces(textposition='top center')
fig.update_yaxes(visible=False)
fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[-1]))
有人对在第一种情况下躲避点或在第二种情况下显示标签有任何想法或解决方法吗？
提前致谢。
PS：这是我为生成图而制作的随机假数据框：
df = pd.DataFrame('coef': 0: 1.0018729737113143,
1: 0.9408864645423858,
2: 0.29796556981484884,
3: -0.6844053575764955,
4: -0.13689631932690113,
5: 0.1473096200402363,
6: 0.9564712505670716,
7: 0.956099003887811,
8: 0.33319108930207175,
9: -0.7022778825729681,
10: -0.1773916842612131,
11: 0.09485417304851751,
'index': 0: 'const',
1: 'x1',
2: 'x2',
3: 'x3',
4: 'x4',
5: 'x5',
6: 'const',
7: 'x1',
8: 'x2',
9: 'x3',
10: 'x4',
11: 'x5',
'label': 0: '1.002***',
1: '0.941***',
2: '0.298***',
3: '-0.684***',
4: '-0.137',
5: '0.147',
6: '0.956***',
7: '0.956***',
8: '0.333***',
9: '-0.702***',
10: '-0.177',
11: '0.095',
'lerr': 0: 0.19788416996400904,
1: 0.19972987383410545,
2: 0.0606849959013587,
3: 0.1772734289533593,
4: 0.1988122854078155,
5: 0.21870366703236832,
6: 0.2734783191688098,
7: 0.2760291042678362,
8: 0.08386739920069491,
9: 0.2449940255063039,
10: 0.27476098595116555,
11: 0.3022511162310027,
'lower': 0: 0.8039888037473053,
1: 0.7411565907082803,
2: 0.23728057391349014,
3: -0.8616787865298547,
4: -0.33570860473471664,
5: -0.07139404699213203,
6: 0.6829929313982618,
7: 0.6800698996199748,
8: 0.24932369010137684,
9: -0.947271908079272,
10: -0.45215267021237865,
11: -0.2073969431824852,
'model': 0: 'OLS',
1: 'OLS',
2: 'OLS',
3: 'OLS',
4: 'OLS',
5: 'OLS',
6: 'QuantReg',
7: 'QuantReg',
8: 'QuantReg',
9: 'QuantReg',
10: 'QuantReg',
11: 'QuantReg',
'pvalue': 0: 1.4211692095019375e-16,
1: 4.3583690618389965e-15,
2: 6.278403727223468e-16,
3: 1.596372747840846e-11,
4: 0.17483151363955116,
5: 0.18433051296752084,
6: 4.877385844808361e-10,
7: 6.665860891682504e-10,
8: 5.476882838731488e-12,
9: 1.4240852942202845e-07,
10: 0.20303143985022934,
11: 0.5347222575215599,
'uerr': 0: 0.19788416996400904,
1: 0.19972987383410556,
2: 0.06068499590135873,
3: 0.1772734289533593,
4: 0.19881228540781554,
5: 0.21870366703236832,
6: 0.27347831916880994,
7: 0.2760291042678362,
8: 0.08386739920069491,
9: 0.2449940255063039,
10: 0.27476098595116555,
11: 0.3022511162310027,
'upper': 0: 1.1997571436753234,
1: 1.1406163383764913,
2: 0.35865056571620757,
3: -0.5071319286231362,
4: 0.0619159660809144,
5: 0.3660132870726046,
6: 1.2299495697358815,
7: 1.2321281081556472,
8: 0.41705848850276667,
9: -0.4572838570666642,
10: 0.09736930168995245,
11: 0.3971052892795202)
【问题讨论】：
【参考方案1】：您的第二次尝试非常接近可行的解决方案。只需为您的标签腾出更多空间：
height=600, width=800
然后将名为'OLS' 的轨迹的标签放置在每个子图的边界内：
fig.for_each_trace(lambda t: t.update(textposition='bottom center') if t.name == 'OLS' else ())
剧情：
完整代码：
import plotly.express as px
import pandas as pd
df = pd.DataFrame('coef': 0: 1.0018729737113143,
1: 0.9408864645423858,
2: 0.29796556981484884,
3: -0.6844053575764955,
4: -0.13689631932690113,
5: 0.1473096200402363,
6: 0.9564712505670716,
7: 0.956099003887811,
8: 0.33319108930207175,
9: -0.7022778825729681,
10: -0.1773916842612131,
11: 0.09485417304851751,
'index': 0: 'const',
1: 'x1',
2: 'x2',
3: 'x3',
4: 'x4',
5: 'x5',
6: 'const',
7: 'x1',
8: 'x2',
9: 'x3',
10: 'x4',
11: 'x5',
'label': 0: '1.002***',
1: '0.941***',
2: '0.298***',
3: '-0.684***',
4: '-0.137',
5: '0.147',
6: '0.956***',
7: '0.956***',
8: '0.333***',
9: '-0.702***',
10: '-0.177',
11: '0.095',
'lerr': 0: 0.19788416996400904,
1: 0.19972987383410545,
2: 0.0606849959013587,
3: 0.1772734289533593,
4: 0.1988122854078155,
5: 0.21870366703236832,
6: 0.2734783191688098,
7: 0.2760291042678362,
8: 0.08386739920069491,
9: 0.2449940255063039,
10: 0.27476098595116555,
11: 0.3022511162310027,
'lower': 0: 0.8039888037473053,
1: 0.7411565907082803,
2: 0.23728057391349014,
3: -0.8616787865298547,
4: -0.33570860473471664,
5: -0.07139404699213203,
6: 0.6829929313982618,
7: 0.6800698996199748,
8: 0.24932369010137684,
9: -0.947271908079272,
10: -0.45215267021237865,
11: -0.2073969431824852,
'model': 0: 'OLS',
1: 'OLS',
2: 'OLS',
3: 'OLS',
4: 'OLS',
5: 'OLS',
6: 'QuantReg',
7: 'QuantReg',
8: 'QuantReg',
9: 'QuantReg',
10: 'QuantReg',
11: 'QuantReg',
'pvalue': 0: 1.4211692095019375e-16,
1: 4.3583690618389965e-15,
2: 6.278403727223468e-16,
3: 1.596372747840846e-11,
4: 0.17483151363955116,
5: 0.18433051296752084,
6: 4.877385844808361e-10,
7: 6.665860891682504e-10,
8: 5.476882838731488e-12,
9: 1.4240852942202845e-07,
10: 0.20303143985022934,
11: 0.5347222575215599,
'uerr': 0: 0.19788416996400904,
1: 0.19972987383410556,
2: 0.06068499590135873,
3: 0.1772734289533593,
4: 0.19881228540781554,
5: 0.21870366703236832,
6: 0.27347831916880994,
7: 0.2760291042678362,
8: 0.08386739920069491,
9: 0.2449940255063039,
10: 0.27476098595116555,
11: 0.3022511162310027,
'upper': 0: 1.1997571436753234,
1: 1.1406163383764913,
2: 0.35865056571620757,
3: -0.5071319286231362,
4: 0.0619159660809144,
5: 0.3660132870726046,
6: 1.2299495697358815,
7: 1.2321281081556472,
8: 0.41705848850276667,
9: -0.4572838570666642,
10: 0.09736930168995245,
11: 0.3971052892795202)
fig = px.scatter(
df, y='model', x='coef', text='label', color='model',
facet_row='index',
error_x_minus='lerr', error_x='uerr',
hover_data=['coef', 'pvalue', 'lower', 'upper'],
height=600, width=800,
)
fig.update_traces(textposition='top center')
fig.update_yaxes(visible=False)
fig.for_each_annotation(lambda a: a.update(text=a.text.split("=")[-1]))
fig.for_each_trace(lambda t: t.update(textposition='bottom center') if t.name == 'OLS' else ())
fig.show()
【讨论】：
以上是关于Plotly：在散点图分类轴上躲避重叠点的主要内容，如果未能解决你的问题，请参考以下文章
Plotly 散点图趋势线出现在散点下方。如何让趋势线出现在散点图上？ [Python]
[散点图][Plotly][Python] 如何在散点图中标记心形