如何用Vega实现按分组字段分面的词云？

一、实现分组网格词云 #

Vega本身没有原生分面功能，但可以通过Repeat机制 + 数据过滤实现网格布局的分组词云，步骤如下：

1. 加载GitHub数据 #

import requests
import json

# 替换为你的GitHub数据URL
data_url = "https://raw.githubusercontent.com/your-repo/your-data.json"
raw_data = requests.get(data_url).json()

# 提取所有分组（用于后续Repeat配置）
all_groups = list({item["group"] for item in raw_data})

2. 构建分面Vega Spec #

核心思路是用repeat定义网格的行/列分组，每个子视图通过Signal过滤对应分组的数据，生成独立词云：

# 基础词云Spec模板
base_wordcloud_spec = {
  "$schema": "https://vega.github.io/schema/vega/v5.json",
  "width": 250,
  "height": 250,
  "padding": 10,
  "signals": [{"name": "target_group", "value": ""}],
  "data": [
    {
      "name": "filtered_data",
      "values": raw_data,
      "transform": [{"type": "filter", "expr": "datum.group == target_group"}]
    },
    {
      "name": "wordcloud_data",
      "source": "filtered_data",
      "transform": [
        # 统计词频（保留3个字符以上的词）
        {"type": "countpattern", "field": "text", "case": "lower", "pattern": "[\\w']{3,}", "stopwords": []},
        # 随机旋转角度
        {"type": "formula", "as": "angle", "expr": "random() * 6.283185307179586"},
        # 词重（长词更粗）
        {"type": "formula", "as": "weight", "expr": "if(datum.text.length > 6, 700, 400)"},
        # 生成词云布局
        {"type": "wordcloud", "size": [{"signal": "width"}, {"signal": "height"}], "text": "text", "rotate": "angle", "fontSize": "count", "fontWeight": "weight", "font": "Arial", "padding": 3}
      ]
    }
  ],
  "marks": [
    {
      "type": "text",
      "from": {"data": "wordcloud_data"},
      "encode": {
        "enter": {
          "text": {"field": "text"},
          "align": {"value": "center"},
          "baseline": {"value": "alphabetic"},
          "fill": {"value": "#2c3e50"}
        },
        "update": {
          "x": {"field": "x"},
          "y": {"field": "y"},
          "rotate": {"field": "rotate"},
          "fontSize": {"field": "fontSize"},
          "fontWeight": {"field": "fontWeight"},
          "fillOpacity": {"value": 1}
        },
        "hover": {"fillOpacity": {"value": 0.6}}
      }
    }
  ]
}

# 构建分面网格Spec（按行排列分组，可改为column或同时row+column）
faceted_spec = {
  "$schema": "https://vega.github.io/schema/vega/v5.json",
  "repeat": {"row": all_groups},  # 改为"column": all_groups则按列排列
  "spec": base_wordcloud_spec,
  "resolve": {"signal": {"target_group": {"repeat": "row"}}}  # 绑定Repeat变量到过滤Signal
}

3. 在Colab中渲染 #

用你熟悉的两种方式都可以：

方式一：Altair Vega

from altair import vega
vega.renderers.enable('colab')
vega.Vega(faceted_spec)

方式二：Panel

import panel as pn
from vega import Vega
pn.extension('vega')
pn.pane.Vega(faceted_spec)

二、在Vega Spec中引用Python变量 #

直接修改Spec的字典结构即可，无需复杂模板：

# 示例：动态设置词云尺寸、字体
custom_width = 300
custom_height = 300
custom_font = "Helvetica Neue"

# 修改基础Spec的参数
base_wordcloud_spec["width"] = custom_width
base_wordcloud_spec["height"] = custom_height
base_wordcloud_spec["data"][1]["transform"][-1]["font"] = custom_font
base_wordcloud_spec["data"][1]["transform"][-1]["size"] = [custom_width, custom_height]

# 重新构建分面Spec（或直接修改已有的faceted_spec）
faceted_spec["spec"] = base_wordcloud_spec

如果需要更复杂的动态逻辑（比如动态选择分组），可以将Spec转为字符串后用Python格式化，再转回JSON：

import json

# 将Spec转为字符串
spec_str = json.dumps(base_wordcloud_spec)
# 替换变量
spec_str = spec_str.replace('"font": "Arial"', f'"font": "{custom_font}"')
# 转回JSON对象
updated_spec = json.loads(spec_str)

内容的提问来源于stack exchange，提问作者sisyphos