如何在Power BI的Vega-Lite(Deneb)中实现柱内抖动散点图？

核心思路 #

要实现每个售出商品在对应柱的营收范围内随机分布的散点，需关联明细数据与品类汇总数据，通过数据转换计算散点坐标偏移；同时基于营收排名实现畅销品类的高亮。

步骤1：数据关联与散点坐标计算 #

利用Vega-Lite的transform完成明细数据与汇总数据的关联，为每个商品计算散点的抖动位置：

• 通过lookup将明细数据与品类汇总表关联，获取对应品类的总营收
• 计算散点的Y轴坐标：在0到品类总营收范围内随机取值（random() * datum.total_revenue）
• 计算散点的X轴抖动偏移：在柱形宽度范围内左右随机偏移（避免超出柱形）

示例transform配置：

"transform": [
  {
    "lookup": "Category",
    "from": {
      "data": {"name": "summary"},
      "key": "Category",
      "fields": ["TotalRevenue"]
    },
    "as": ["total_revenue"]
  },
  {
    "calculate": "datum.Category + (random() - 0.5) * 0.7",
    "as": "scatter_x"
  },
  {
    "calculate": "random() * datum.total_revenue",
    "as": "scatter_y"
  },
  // 计算营收排名，用于高亮畅销品类
  {
    "window": [{"op": "rank", "field": "total_revenue", "as": "revenue_rank"}],
    "sort": [{"field": "total_revenue", "order": "descending"}]
  }
]

步骤2：图层组合（柱形+散点） #

使用Vega-Lite的layer将柱形图与散点图叠加：

• 底层为品类汇总柱形图，设置基础样式
• 上层为抖动散点图，使用计算出的scatter_x和scatter_y作为坐标

步骤3：畅销品类高亮 #

通过condition编码实现柱形、散点及轴标签的高亮：

• 柱形与散点：判断revenue_rank === 1时使用高亮色
• X轴标签：同样通过condition为畅销品类设置加粗/变色样式

完整示例代码 #

{
  "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
  "data": {"name": "dataset"},
  "datasets": {
    "summary": [
      // 替换为你的汇总数据，或直接关联Power BI的汇总表
      {"Category": "品类A", "TotalRevenue": 15000},
      {"Category": "品类B", "TotalRevenue": 22000},
      {"Category": "品类C", "TotalRevenue": 18000}
    ]
  },
  "transform": [
    {
      "lookup": "Category",
      "from": {
        "data": {"name": "summary"},
        "key": "Category",
        "fields": ["TotalRevenue"]
      },
      "as": ["total_revenue"]
    },
    {
      "calculate": "datum.Category + (random() - 0.5) * 0.7",
      "as": "scatter_x"
    },
    {
      "calculate": "random() * datum.total_revenue",
      "as": "scatter_y"
    },
    {
      "window": [{"op": "rank", "field": "total_revenue", "as": "revenue_rank"}],
      "sort": [{"field": "total_revenue", "order": "descending"}]
    }
  ],
  "layer": [
    // 柱形图层
    {
      "mark": {"type": "bar", "width": 0.8},
      "encoding": {
        "x": {"field": "Category", "type": "nominal"},
        "y": {"field": "total_revenue", "type": "quantitative"},
        "color": {
          "condition": {"test": "datum.revenue_rank === 1", "value": "#e63946"},
          "value": "#a8dadc"
        }
      }
    },
    // 散点图层
    {
      "mark": {"type": "point", "size": 30, "opacity": 0.6},
      "encoding": {
        "x": {"field": "scatter_x", "type": "nominal", "axis": null},
        "y": {"field": "scatter_y", "type": "quantitative", "axis": null},
        "color": {
          "condition": {"test": "datum.revenue_rank === 1", "value": "#1d3557"},
          "value": "#457b9d"
        }
      }
    }
  ],
  "config": {
    "axis": {
      "x": {
        "labelFontWeight": {
          "condition": {"test": "datum.revenue_rank === 1", "value": "bold"},
          "value": "normal"
        },
        "labelColor": {
          "condition": {"test": "datum.revenue_rank === 1", "value": "#e63946"},
          "value": "#333"
        }
      }
    }
  }
}

关键说明 #

• 若使用Power Query转换的单条商品行数据，直接将dataset关联为该数据源即可
• 抖动范围（0.7）可根据柱宽调整，避免散点超出柱形边界
• 随机数基于每个明细行计算，确保每个商品对应唯一散点

内容的提问来源于stack exchange，提问作者PBI DataVizzle