绘制散点图 - 如何防止标记文本重叠？

Question

我尝试使用 Scattergeo 绘制美国地图。我有很多彼此靠近的标记，它们的标记文本相互重叠，看起来很混乱。有解决方法吗？我正在努力使地图看起来尽可能干净。

我不一定需要标记位于其精确坐标，但我不想手动调整每个地理坐标，因为坐标太多。所以程序化的东西是最好的。

Answer 1

这是一个棘手的问题。除了缩小字体的明显想法之外，一种可能性是创建一些按钮，其中每个按钮切换相同比例的标记文本标签。如果您有 50 个标记文本标签，则可以使用 10 个按钮，每个按钮切换 5 个标记文本标签，从而减少它们重叠的机会。您可以相应地调整按钮的数量。

下面是一个包含机场数据的示例，否则会有很多重叠的文本：

import numpy as np
import pandas as pd
import plotly.graph_objects as go

df = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/2011_february_us_airport_traffic.csv')
df['text'] = df['airport'] + '' + df['city'] + ', ' + df['state'] + '' + 'Arrivals: ' + df['cnt'].astype(str)

## assign clusters to data at random
np.random.seed(42)
n_clusters = 10
cluster_choices = [f'Cluster_{i}' for i in range(n_clusters)]
df['cluster'] = np.random.choice(cluster_choices, size=len(df))

fig = go.Figure()

## only show the text for first cluster of markers
for group, df_group in df.groupby('cluster'):
    if group == 'Cluster_1':
        print("cluster 1 is visible")
        visible = True
    else:
        visible = False
    
    fig.add_trace(go.Scattergeo(
        lon = df_group['long'],
        lat = df_group['lat'],
        text = df_group['airport'],
        mode = 'text',
        visible = visible
    ))

## always show markers
fig.add_trace(go.Scattergeo(
    lon = df['long'],
    lat = df['lat'],
    text = df['airport'],
    mode = 'markers',
    marker_color = df['cnt'],
    visible = True,
))

## there are n_clusters text traces, followed by n_clusters marker traces
## so we always show the last n_clusters out of all traces
visible_array = [[i == j for j in range(n_clusters)] + [True]*n_clusters for i in range(n_clusters)]

updatemenus = [
    dict(
        buttons=[
            dict(
                label=f'Cluster {i}',
                method='update',
                args=[{'visible': visible_array[i]},
                      {'title': f'Cluster {i}'}]
            ) for i in range(n_clusters)
        ]
    )
]

fig.update_layout(
    title = 'Most trafficked US airports<br>(Hover for airport names)',
    geo_scope='usa',
    showlegend=False,
    updatemenus=updatemenus
)

fig.show()