Show HN: 我们为实时仪表盘构建了自定义图表类型
1 分•作者: YannickCrabbe•8 个月前
大多数分析工具会提供 10-15 种图表类型，然后就完事了。但如果这些图表都无法很好地呈现你的数据，该怎么办？
我们在构建 CRM 仪表板时就遇到了这个问题。我们想展示销售代表如何与未结交易建立联系：不仅仅是总数，而是实际的关系。条形图太扁平，表格又太密集。我们需要一个网络图。
问题是？大多数图表库和 BI 工具都不支持这种图表。你只能对功能请求说“不”，或者拼凑一个独立的视觉效果，而这会破坏产品的流程（没有筛选器，没有交互性，没有主题）。
所以我们自己构建了它，但它仍然可以很好地与仪表板的其余部分配合使用。现在，Luzmo 允许你定义自己的图表类型，编写自己的可视化代码，然后像其他任何图表一样将其放入仪表板编辑器中。Luzmo 仍然处理那些无聊的事情：查询、筛选、主题和跨图表链接。
最终结果：
- 销售代表成为网络图中的节点，未结交易围绕着他们。
- 交易规模控制节点大小；获胜概率控制颜色。
- 所有内容都响应筛选器，并与开箱即用的其他图表进行交互。
教程和 GitHub 存储库将逐步介绍整个过程：设置构建器、定义数据槽、编写渲染方法，以及打包所有内容以进行部署。
我很乐意听取其他解决过类似问题的人的意见：
- 你是否曾经需要 BI 工具不支持的图表类型？
- 你是自己构建的，还是找到了解决方法？
- 原生交互性与仅仅嵌入一个独立的视觉效果相比，有多重要？
阅读文章 + 查看完整代码：<a href="https://www.luzmo.com/blog/build-custom-charts" rel="nofollow">https://www.luzmo.com/blog/build-custom-charts</a>
查看原文
Most analytics tools give you 10–15 chart types and call it a day. But what happens when none of them actually represent your data well?<p>We ran into this when building a CRM dashboard. We wanted to show how sales reps connect to open deals: not just totals, but actual relationships. A bar chart was too flat, a table too dense. We needed a network graph.<p>The problem? Most chart libraries and BI tools don’t support that out of the box. You either say “no” to the feature request, or you hack together a standalone visualization that breaks the product’s flow (no filters, no interactivity, no theming).<p>So we built it ourselves, but in a way that still plays nicely with the rest of the dashboard. Luzmo now lets you define your own chart type, write your own visualization code, and then drop it into the dashboard editor like any other chart. Luzmo still handles the boring stuff: querying, filtering, theming, and cross-chart linking.<p>The end result:
- Sales reps become nodes in a network graph, with open deals orbiting around them.
- Deal size controls node size; win probability controls color.
- Everything responds to filters and interacts with other charts out of the box.<p>The tutorial and GitHub repo walk through the whole process: setting up the builder, defining data slots, writing render methods, and packaging it all up for deployment.<p>I’d love to hear from others who’ve solved similar problems:
- Have you ever needed a chart type your BI tool didn’t support?
- Did you build it yourself, or find a workaround?
- How important is native interactivity vs. just embedding a standalone visualization?<p>Read the post + see the full code: <a href="https://www.luzmo.com/blog/build-custom-charts" rel="nofollow">https://www.luzmo.com/blog/build-custom-charts</a>