如何在Python的Seaborn Pointplot中分组连接点并添加分类颜色

我正在用Python开展项目，在Kaggle上看到一幅用R绘制的精美图表，但我不熟悉R语言。测试数据采用Kaggle的【Monthly Electricity Production in GWh [2010-2022]】，目前已用Python的Seaborn库复刻出接近的图表，现有代码如下：

grid = sns.FacetGrid(data=breakdown, col="country", col_wrap=7, height=3, aspect=1, sharex=True, sharey=False)
grid.map_dataframe(sns.pointplot, x="value", y="product", hue="year", join=False, markers=["o", "*"])
grid.set_titles("{col_name}")
grid.set_axis_labels("[%]", "")
grid.set(xlim=(0, 100), ylim=(-0.5, 2.5))
grid.despine(right=True, top=True)
grid.add_legend(loc="upper left", frameon=False)

现在需要实现两个需求：

• 同组（同一国家+同一能源类型）内不同年份的点用线连接
• 为Nuclear、Renewables、Fossil fuels三个能源分类设置对应固定颜色

所用breakdown数据集样本：

,country,year,product,value
0,Australia,2010,Renewables,10.400785746199682
12,Australia,2022,Renewables,34.742331289873064
467,Australia,2022,Nuclear,0.0
455,Australia,2010,Nuclear,0.0
910,Australia,2010,Fossil fuels,89.59921045031392
922,Australia,2022,Fossil fuels,65.25766871012695
13,Austria,2010,Renewables,67.75576873959946
25,Austria,2022,Renewables,78.76446322617821
480,Austria,2022,Nuclear,0.0
468,Austria,2010,Nuclear,0.0
923,Austria,2010,Fossil fuels,32.24423569765363
935,Austria,2022,Fossil fuels,21.235536776894953
493,Belgium,2022,Nuclear,45.295925696472636
481,Belgium,2010,Nuclear,49.984697658490894
948,Belgium,2022,Fossil fuels,27.7838958679145
38,Belgium,2022,Renewables,26.9201784290824
26,Belgium,2010,Renewables,8.396529615422356
936,Belgium,2010,Fossil fuels,41.61877381915019
494,Canada,2010,Nuclear,14.608937138869305
949,Canada,2010,Fossil fuels,22.769913684560386
961,Canada,2022,Fossil fuels,17.25003004705112
51,Canada,2022,Renewables,69.86121900236432
39,Canada,2010,Renewables,62.62114900575948
506,Canada,2022,Nuclear,12.888750950897817

解决方案 #

1. 定义能源分类的颜色映射

先创建一个字典，为每个能源类型指定固定颜色，可根据需求调整颜色值：

color_map = {
    "Nuclear": "#27ae60",    # 绿色
    "Renewables": "#3498db", # 蓝色
    "Fossil fuels": "#e74c3c"# 红色
}

2. 修改Seaborn绘图代码

核心调整点：

• 将hue参数从year改为product，让同能源类型的点自动连接
• 用style参数区分不同年份，保留原有的标记样式
• 指定palette为自定义颜色映射
• 调整图例显示，避免重复项

修改后的完整代码：

# 先对数据排序，确保连线顺序按年份排列
breakdown_sorted = breakdown.sort_values(by=["country", "product", "year"])

grid = sns.FacetGrid(data=breakdown_sorted, col="country", col_wrap=7, height=3, aspect=1, sharex=True, sharey=False)
# 使用pointplot实现同能源类型点的连接，用style区分年份
grid.map_dataframe(
    sns.pointplot,
    x="value",
    y="product",
    hue="product",
    style="year",
    join=True,
    markers=["o", "*"],
    palette=color_map,
    dodge=False  # 关闭偏移，让同能源类型的点在同一y轴位置对齐
)
grid.set_titles("{col_name}")
grid.set_axis_labels("[%]", "")
grid.set(xlim=(0, 100), ylim=(-0.5, 2.5))
grid.despine(right=True, top=True)

# 调整图例：去重并统一位置
handles, labels = grid.axes[0].get_legend_handles_labels()
filtered_handles = []
filtered_labels = []
seen = set()
for h, l in zip(handles, labels):
    if l not in seen:
        seen.add(l)
        filtered_handles.append(h)
        filtered_labels.append(l)
grid.fig.legend(filtered_handles, filtered_labels, loc="upper left", frameon=False, bbox_to_anchor=(0.05, 0.95))
# 移除子图默认图例
for ax in grid.axes.flat:
    ax.legend_.remove()

说明

• 排序数据是为了保证连线按2010到2022的年份顺序，避免连线混乱
• dodge=False让同一能源类型的不同年份点在同一y轴位置对齐，连线更直观
• 图例去重处理避免了每个子图重复显示图例，让整体布局更整洁

内容的提问来源于stack exchange，提问作者Gabriel Facheti