interactive-graphics.qmd

# 交互图形 {#sec-interactive-graphics}

```{r}
#| echo: false

source("_common.R")
```

```{css}
#| echo: false

.modebar {
  display: none !important;
}
```

在之前的数据探索章节介绍了 **ggplot2** 包，本章将介绍 **plotly** 包，绘制交互图形，包含基础元素、常用图形和技巧，沿用日志提交数据和 Base R 内置的斐济及周边地震数据。写作上，仍然以一个数据串联尽可能多的小节，从 **ggplot2** 包到 **plotly** 包，将介绍其间的诸多联系，以便读者轻松掌握。

## 基础元素 {#sec-interactive-elements}

### 图层 {#sec-plotly-layer}

**plotly** 包封装了许多图层函数，可以绘制各种各样的统计图形，见下 @tbl-plotly-layers 。

```{r}
#| eval: false
#| echo: false

library(plotly, warn.conflicts = FALSE)
knitr::kable(matrix(grep("add_*", x = ls("package:plotly"), value = T), ncol = 3),
  col.names = c("A", "B", "C"), caption = "图层", format = "pipe"
)
```

|                 |                        |                |
|:----------------|:-----------------------|:---------------|
| add_annotations | add_histogram          | add_polygons   |
| add_area        | add_histogram2d        | add_ribbons    |
| add_bars        | add_histogram2dcontour | add_scattergeo |
| add_boxplot     | add_image              | add_segments   |
| add_choropleth  | add_lines              | add_sf         |
| add_contour     | add_markers            | add_surface    |
| add_data        | add_mesh               | add_table      |
| add_fun         | add_paths              | add_text       |
| add_heatmap     | add_pie                | add_trace      |

: **plotly** 包可以绘制丰富的统计图形 {#tbl-plotly-layers}

下面以散点图为例，使用方式非常类似 **ggplot2** 包，函数 `plot_ly()` 类似 `ggplot()`，而函数 `add_markers()` 类似 `geom_point()`，效果如 @fig-plotly-markers 所示。

```{r}
#| label: fig-plotly-markers-html
#| eval: false
#| echo: true

# https://plotly.com/r/reference/scatter/
plotly::plot_ly(data = quakes, x = ~long, y = ~lat) |> 
  plotly::add_markers()
```

```{r}
#| label: fig-plotly-markers
#| fig-cap: "默认风格的简单散点图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-formula.png")
```

或者使用函数 `add_trace()`，层层添加图形元素，效果和上 @fig-plotly-markers 是一样的。

```{r}
#| label: fig-plotly-trace
#| eval: false

plotly::plot_ly(data = quakes, x = ~long, y = ~lat) |> 
  plotly::add_trace(type = "scatter", mode = "markers")
```

::: callout-tip
**plotly** 包的函数 `plot_ly()` 又与 **ggplot2** 包中函数 `qplot()` 类似，可以将大部分设置塞进去。

```{r}
#| label: fig-plotly-qplot
#| eval: false

plotly::plot_ly(
  data = quakes, x = ~long, y = ~lat,
  type = "scatter", mode = "markers"
)
```

所以，总的来说， `add_markers()` 、`add_trace(type = "scatter", mode = "markers")` 和 `plot_ly(type = "scatter", mode = "markers")` 是等价的。
:::

### 配色 {#sec-plotly-color}

在 @fig-plotly-markers 的基础上，将颜色映射到震级变量上。

```{r}
#| label: fig-plotly-color
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "给散点图配色"

plotly::plot_ly(data = quakes, x = ~long, y = ~lat) |> 
  plotly::add_markers(color = ~mag)
```

### 刻度 {#sec-plotly-scale}

东经和南纬

```{r}
#| label: fig-plotly-scale
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "设置刻度及标签"

plotly::plot_ly(data = quakes, x = ~long, y = ~lat) |> 
  plotly::add_markers(color = ~mag) |>
  plotly::layout(
    xaxis = list(title = "经度", ticksuffix = 'E'),
    yaxis = list(title = "纬度", ticksuffix = 'S')
  )
```

### 标签 {#sec-plotly-label}

添加横轴、纵轴以及主副标题

```{r}
#| label: fig-plotly-label
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "添加各处标题"

plotly::plot_ly(
  data = quakes, x = ~long, y = ~lat,
  marker = list(
    color = ~mag,
    colorscale = "Viridis",
    colorbar = list(title = list(text = "震级"))
  )
) |>
  plotly::add_markers() |>
  plotly::layout(
    xaxis = list(title = "经度"),
    yaxis = list(title = "纬度"),
    title = "斐济及其周边地区的地震活动"
  )
```

### 主题 {#sec-plotly-theme}

plotly 内置了一些主题风格

```{r}
#| label: fig-plotly-theme
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "设置主题风格"

plotly::plot_ly(
  data = quakes, x = ~long, y = ~lat,
  marker = list(
    color = ~mag,
    colorscale = "Viridis",
    colorbar = list(title = list(text = "震级"))
  )
) |>
  plotly::add_markers() |>
  plotly::layout(
    xaxis = list(title = "经度"),
    yaxis = list(title = "纬度"),
    title = "斐济及其周边地区的地震活动"
  )
```

### 字体 {#sec-plotly-font}

### 图例 {#sec-plotly-legend}

## 常用图形 {#sec-plotly-common-graphics}

### 散点图 {#sec-plotly-scatter}

**plotly** 包支持绘制许多常见的散点图，从直角坐标系 `scatter` 到极坐标系 `scatterpolar` 和地理坐标系 `scattergeo`，从二维平面 `scatter` 到三维空间 `scatter3d`，借助 WebGL 可以渲染大规模的数据点 `scattergl`。

| 类型             | 名称                          |
|:-----------------|:------------------------------|
| `scatter`        | 二维平面散点图                |
| `scatter3d`      | 三维立体散点图                |
| `scattergl`      | 散点图（WebGL 版）            |
| `scatterpolar`   | 极坐标下散点图                |
| `scatterpolargl` | 极坐标下散点图（WebGL 版）    |
| `scattergeo`     | 地理坐标下散点图              |
| `scattermapbox`  | 地理坐标下散点图（MapBox 版） |
| `scattercarpet`  | 地毯图                        |
| `scatterternary` | 三元图                        |

: plotly 包支持绘制的散点图类型 {#tbl-plotly-scatter}

@fig-scatter 展示斐济及其周边的地震分布

```{r}
#| label: fig-scatter-html
#| eval: false
#| echo: true

plotly::plot_ly(
  data = quakes, x = ~long, y = ~lat,
  type = "scatter", mode = "markers"
) |>
  plotly::layout(
    xaxis = list(title = "经度"),
    yaxis = list(title = "纬度")
  )
```

```{r}
#| label: fig-scatter
#| fig-cap: "普通散点图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-scatter.png")
```

### 柱形图 {#sec-plotly-bar}

```{r}
#| echo: false

svn_trunk_log <- readRDS(file = "data/svn-trunk-log-2022.rds")
svn_trunk_log <- within(svn_trunk_log, {
  # 提取日期、月份、年份、星期、第几周、第几天等时间成分
  year <- as.integer(format(stamp, "%Y"))
  date <- format(stamp, format = "%Y-%m-%d", tz = "UTC")
  month <- format(stamp, format = "%m", tz = "UTC")
  hour <- format(stamp, format = "%H", tz = "UTC")
  week <- format(stamp, format = "%U", tz = "UTC")
  wday <- format(stamp, format = "%a", tz = "UTC")
  nday <- format(stamp, format = "%j", tz = "UTC")
})
trunk_year <- aggregate(data = svn_trunk_log, revision ~ year, FUN = length)
```

```{r}
#| label: fig-plotly-bar
#| fig-cap: "柱形图"
#| eval: !expr knitr::is_html_output(excludes = 'epub')

# https://plotly.com/r/reference/bar/
plotly::plot_ly(
  data = trunk_year, x = ~year, y = ~revision, type = "bar"
) |>
  plotly::layout(
    xaxis = list(title = "年份"),
    yaxis = list(title = "代码提交量")
  )
```

### 曲线图 {#sec-plotly-line}

```{r}
#| label: fig-plotly-curve
#| fig-cap: "曲线图"
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::plot_ly(
  data = trunk_year, x = ~year, y = ~revision, type = "scatter",
  mode = "markers+lines", line = list(shape = "spline")
) |> 
  plotly::layout(
    xaxis = list(title = "年份"),
    yaxis = list(title = "代码提交量")
  )
```

### 直方图 {#sec-plotly-hist}

地震次数随震级的分布变化，下 @fig-quakes-mag 为频数分布图

```{r}
#| label: fig-quakes-mag-html
#| eval: false
#| echo: true

# https://plotly.com/r/reference/histogram/
plotly::plot_ly(quakes, x = ~mag, type = "histogram") |> 
  plotly::layout(
    xaxis = list(title = "震级"),
    yaxis = list(title = "次数")
  )
```

```{r}
#| label: fig-quakes-mag
#| fig-cap: "地震震级的频数分布图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-quakes-mag.png")
```

地震震级的概率分布，下 @fig-quakes-mag-prob 为频率分布图

```{r}
#| label: fig-quakes-mag-prob-html
#| eval: false
#| echo: true

plotly::plot_ly(
  data = quakes, x = ~mag, type = "histogram",
  histnorm = "probability",
  marker = list(
    color = "lightblue",
    line = list(color = "white", width = 2)
  )
) |>
  plotly::layout(
    xaxis = list(title = "震级"),
    yaxis = list(title = "频率")
  )
```

```{r}
#| label: fig-quakes-mag-prob
#| fig-cap: "地震震级的频率分布图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-quakes-mag-prob.png")
```

`histnorm = "probability"` 意味着纵轴表示频率，即每个窗宽下地震次数占总地震次数的比例。地震常常发生在地下，不同的深度对应着不同的地质构造、不同的地震成因，下 @fig-quakes-depth-hist 展示海平面下不同深度的地震震级分布。

```{r}
quakes$depth_bin <- cut(quakes$depth, breaks = 150 * 0:5)
```

```{r}
#| label: fig-quakes-depth-hist-html
#| eval: false
#| echo: true

plotly::plot_ly(quakes,
  x = ~mag, colors = "viridis",
  color = ~depth_bin, type = "histogram"
) |> 
  plotly::layout(
    xaxis = list(title = "震级"),
    yaxis = list(title = "次数")
  )
```

```{r}
#| label: fig-quakes-depth-hist
#| fig-cap: "地震震级的频率分布图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-quakes-depth-hist.png")
```

### 箱线图 {#sec-plotly-boxplot}

```{r}
#| label: fig-quakes-depth-box
#| fig-cap: "不同深度下地震震级的分布"
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::plot_ly(quakes,
  x = ~depth_bin, y = ~mag, colors = "viridis",
  color = ~depth_bin, type = "box"
) |> 
  plotly::layout(
    xaxis = list(title = "深度"),
    yaxis = list(title = "震级")
  )
```

```{r}
#| label: fig-quakes-depth-violin
#| fig-cap: "不同深度下地震震级的分布"
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::plot_ly(quakes,
  x = ~depth_bin, y = ~mag, split = ~depth_bin,
  type = "violin", color = ~depth_bin, colors = "viridis",
  box = list(visible = TRUE),
  meanline = list(visible = TRUE)
) |> 
  plotly::layout(
    xaxis = list(title = "深度"),
    yaxis = list(title = "震级")
  )
```

### 热力图 {#sec-plotly-densitymap}

plotly 整合了开源的 [Mapbox GL JS](https://github.com/mapbox/mapbox-gl-js)，可以使用 Mapbox 提供的瓦片地图服务（Mapbox Tile Maps），对空间点数据做核密度估计，展示热力分布，如 @fig-densitymap 所示。图左上角为所罗门群岛（Solomon Islands）、瓦努阿图（Vanuatu）和新喀里多尼亚（New Caledonia），图下方为新西兰北部的威灵顿（Wellington）和奥克兰（Auckland），图中部为斐济（Fiji）。

```{r}
#| label: fig-densitymap-html
#| eval: false
#| echo: true

plotly::plot_ly(
  data = quakes, lat = ~lat, lon = ~long, radius = 10,
  type = "densitymapbox", coloraxis = "coloraxis"
) |>
  plotly::layout(
    mapbox = list(
      style = "stamen-terrain", zoom = 3,
      center = list(lon = 180, lat = -25)
    ),
    coloraxis = list(colorscale = "Viridis")
  )
```

```{r}
#| label: fig-densitymap
#| fig-cap: "空间点数据的核密度估计"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-densitymap.png")
```

图中设置瓦片地图的风格 `style` 为 `"stamen-terrain"`，还可以使用其他开放的栅格瓦片地图服务，比如 `"open-street-map"` 和 `"carto-positron"`。如果使用 MapBox 提供的矢量瓦片地图服务，则需要访问令牌 Mapbox Access Token。图中设置中心坐标 `center` 以及缩放倍数 `zoom`，目的是突出图片中的数据区域。设置调色板 Viridis 展示热力分布，黄色团块的地方表示地震频次高。

### 面量图 {#sec-plotly-choropleth-maps}

在之前我们介绍过用 ggplot2 绘制地区分布图，实际上，地区分布图还有别名，如围栏图、面量图等。本节使用 plotly 绘制交互式的地区分布图，如 @fig-choropleth-map 所示。

```{r}
#| label: fig-choropleth-map-html
#| eval: false
#| echo: true

# https://plotly.com/r/reference/choropleth/
dat <- data.frame(state.x77,
  stats = rownames(state.x77),
  stats_abbr = state.abb
)
# 绘制图形
plotly::plot_ly(
  data = dat,
  type = "choropleth",
  locations = ~stats_abbr,
  locationmode = "USA-states",
  colorscale = "Viridis",
  colorbar = list(title = list(text = "人均收入")),
  z = ~Income
) |>
  plotly::layout(
    geo = list(scope = "usa"),
    title = "1974年美国各州的人均收入"
  )
```

```{r}
#| label: fig-choropleth-map
#| fig-cap: "1974 年美国各州的人均收入"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-choropleth-map.png")
```

### 动态图 {#sec-plotly-animation}

本节参考 plotly 包的官方示例[渐变动画](https://plotly.com/r/cumulative-animations/)，数据来自 SVN 代码提交日志，统计 Martin Maechler 和 Brian Ripley 的年度代码提交量，他们是 R Core Team 非常重要的两位成员，长期参与维护 R 软件及社区。下图展示 1999-2022 年 Martin Maechler 和 Brian Ripley 的代码提交量变化。

```{r}
#| label: fig-plotly-animation
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "1999-2022 年 Martin Maechler 和 Brian Ripley 的代码提交量变化"

# https://plotly.com/r/animations/
trunk_year_author <- aggregate(data = svn_trunk_log, revision ~ year + author, FUN = length)
# https://plotly.com/r/cumulative-animations/
accumulate_by <- function(dat, var) {
  var <- lazyeval::f_eval(f = var, data = dat)
  lvls <- plotly:::getLevels(var) 
  dats <- lapply(seq_along(lvls), function(x) {
    cbind(dat[var %in% lvls[seq(1, x)], ], frame = lvls[[x]])
  })
  dplyr::bind_rows(dats)
}

subset(trunk_year_author, year >= 1999 & author %in% c("ripley", "maechler")) |>
  accumulate_by(~year) |>
  plotly::plot_ly(
    x = ~year, y = ~revision, split = ~author,
    frame = ~frame, type = "scatter", mode = "lines",
    line = list(simplyfy = F)
  ) |>
  plotly::layout(
    xaxis = list(title = "年份"),
    yaxis = list(title = "代码提交量")
  ) |>
  plotly::animation_opts(
    frame = 100, transition = 0, redraw = FALSE
  ) |>
  plotly::animation_button(
    visible = TRUE, # 显示播放按钮
    label = "播放", # 按钮文本
    font = list(color = "gray")# 文本颜色
  ) |>
  plotly::animation_slider(
    currentvalue = list(
      prefix = "年份 ",
      xanchor = "right",
      font = list(color = "gray", size = 30)
    )
  )
```

[**lazyeval**](https://github.com/hadley/lazyeval) 的非标准计算采用 Base R 实现，目前，已经可以被 [**rlang**](https://github.com/r-lib/rlang/) 替代。

## 常用技巧 {#sec-plotly-common-tricks}

### 数学公式 {#sec-plotly-formula}

正态分布的概率密度函数形式如下：

$$
\begin{aligned}
& f(x;\mu,\sigma^2) = \frac{1}{\sqrt{2\pi}\sigma}\exp\{-\frac{(x -\mu)^2}{2\sigma^2}\}
\end{aligned}
$$

下图展示两个正态分布，分别是 $\mathcal{N}(3, 1^2)$ 和 $\mathcal{N}(2, 1.5^2)$ 。函数 `plotly::TeX()` 包裹 LaTeX 书写的数学公式，**plotly** 包调用 [MathJax](https://www.mathjax.org/) 库渲染图中的公式符号。

```{r}
#| eval: false
#| code-fold: true

x <- seq(from = -4, to = 8, length.out = 193)
y1 <- dnorm(x, mean = 3, sd = 1)
y2 <- dnorm(x, mean = 2, sd = 1.5)

plotly::plot_ly(
  x = x, y = y1, type = "scatter", mode = "lines",
  fill = "tozeroy", fillcolor = "rgba(0, 204, 102, 0.2)",
  text = ~ paste0(
    "x：", x, "<br>",
    "y：", round(y1, 3), "<br>"
  ),
  hoverinfo = "text",
  name = plotly::TeX("\\mathcal{N}(3,1^2)"),
  line = list(shape = "spline", color = "#009B95")
) |> 
  plotly::add_trace(
    x = x, y = y2, type = "scatter", mode = "lines",
    fill = "tozeroy", fillcolor = "rgba(51, 102, 204, 0.2)",
    text = ~ paste0(
      "x：", x, "<br>",
      "y：", round(y2, 3), "<br>"
    ),
    hoverinfo = "text",
    name = plotly::TeX("\\mathcal{N}(2, 1.5^2)"),
    line = list(shape = "spline", color = "#403173")
  ) |> 
  plotly::layout(
    xaxis = list(showgrid = F, title = plotly::TeX("x")),
    yaxis = list(showgrid = F, title = plotly::TeX("f(x)")),
    legend = list(x = 0.8, y = 1, orientation = "v")
  ) |> 
  plotly::config(mathjax = "cdn", displayModeBar = FALSE)
```

```{r}
#| label: fig-plotly-formula
#| fig-cap: "设置数学公式"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-formula.png")
```

### 动静转化 {#sec-plotly-ggplotly}

在出版书籍，发表期刊文章，打印纸质文稿等场景中，需要将交互图形导出为静态图形，再插入到正文之中。

```{r}
#| label: fig-ggplot-to-ggplotly
#| fig-cap: "ggplot2 绘制的静态图形"
#| fig-showtext: true

library(ggplot2)
p <- ggplot(data = quakes, aes(x = long, y = lat)) +
  geom_point()
p
```

将 **ggplot2** 包绘制的散点图转化为交互式的散点图，只需调用 **plotly** 包的函数 `ggplotly()`。

```{r}
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::ggplotly(p)
```

当使用配置函数 `config()` 设置参数选项 `staticPlot = TRUE`，可将原本交互式的动态图形转为非交互式的静态图形。

```{r}
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::ggplotly(p) |> 
  plotly::config(staticPlot = TRUE)
```

::: callout-tip
函数 `style()` 设置动态点的注释，比如点横纵坐标、坐标文本，以及整个注释标签的样式，如背景色。

```{r}
#| eval: !expr knitr::is_html_output(excludes = 'epub')

plotly::ggplotly(p, dynamicTicks = "y") |> 
  plotly::style(hoveron = "points", hoverinfo = "x+y+text", 
        hoverlabel = list(bgcolor = "white"))
```
:::

orca (Open-source Report Creator App) 软件针对 **plotly.js** 库渲染的图形具有很强的导出功能，[安装 orca](https://github.com/plotly/orca#installation) 后，`plotly::orca()` 函数可以将基于 **htmlwidgets** 的 **plotly** 图形对象导出为 PNG、PDF 和 SVG 等格式的高质量静态图片。

```{r}
#| eval: false

# orca
plotly::orca(p, "plotly-quakes.svg")
# kaleido
plotly::save_image(p, "plotly-quakes.svg")
```

### 坐标系统 {#sec-plotly-coord}

quakes 是一个包含空间位置的数据集，plotly 的 `scattergeo` 图层 针对空间数据提供多边形矢量边界地图数据，支持设定坐标参考系。下 @fig-scattergeo 增加了地震震级维度，在空间坐标参考系下绘制散点。

```{r}
#| label: fig-scattergeo-html
#| eval: false
#| echo: true

plotly::plot_ly(
  data = quakes,
  lon = ~long, lat = ~lat,
  type = "scattergeo", mode = "markers",
  text = ~ paste0(
    "站点：", stations, "<br>",
    "震级：", mag
  ),
  marker = list(
    color = ~mag, colorscale = "Viridis",
    size = 10, opacity = 0.8,
    line = list(color = "white", width = 1)
  )
) |>
  plotly::layout(geo = list(
    showland = TRUE,
    landcolor = plotly::toRGB("gray95"),
    countrycolor = plotly::toRGB("gray85"),
    subunitcolor = plotly::toRGB("gray85"),
    countrywidth = 0.5,
    subunitwidth = 0.5,
    lonaxis = list(
      showgrid = TRUE,
      gridwidth = 0.5,
      range = c(160, 190),
      dtick = 5
    ),
    lataxis = list(
      showgrid = TRUE,
      gridwidth = 0.5,
      range = c(-40, -10),
      dtick = 5
    )
  ))
```

```{r}
#| label: fig-scattergeo
#| fig-cap: "空间点数据图"
#| out-width: 80%
#| echo: false

knitr::include_graphics(path = "images/plotly-scattergeo.png")
```

### 添加水印 {#sec-plotly-watermark}

在图片右下角添加水印图片

```{r}
#| label: fig-watermark
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "添加水印图片"

plotly::plot_ly(quakes,
  x = ~long, y = ~lat, color = ~mag, 
  type = "scatter", mode = "markers"
) |> 
  plotly::config(staticPlot = TRUE) |> 
  plotly::layout(
    images = list( # 水印图片
      source = "https://images.plot.ly/language-icons/api-home/r-logo.png",
      xref = "paper", # 页面参考
      yref = "paper",
      x = 0.90, # 横坐标
      y = 0.20, # 纵坐标
      sizex = 0.2, # 长度
      sizey = 0.2, # 宽度
      opacity = 0.5 # 透明度
    )
  )
```

### 多图布局 {#sec-plotly-subplot}

将两个图形做上下排列

```{r}
#| label: fig-tagList
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "上下布局"

p1 <- plotly::plot_ly(
  data = trunk_year, x = ~year, y = ~revision, type = "bar"
) |>
  plotly::layout(
    xaxis = list(title = "年份"),
    yaxis = list(title = "代码提交量")
  )

p2 <- plotly::plot_ly(
  data = trunk_year, x = ~year, y = ~revision, type = "scatter",
  mode = "markers+lines", line = list(shape = "spline")
) |>
  plotly::layout(
    xaxis = list(title = "年份"),
    yaxis = list(title = "代码提交量")
  )

htmltools::tagList(p1, p2)
```

plotly 包提供的函数 `subplot()` 专门用于布局排列，下图的上下子图共享 x 轴。

```{r}
#| label: fig-subplot-1
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "上下布局"

plotly::subplot(plotly::style(p1, showlegend = FALSE), 
                plotly::style(p2, showlegend = FALSE), 
                nrows = 2, margin = 0.05, shareX = TRUE, titleY = TRUE)
```

下图展示更加灵活的布局形式，嵌套使用布局函数 `subplot()` 实现。

```{r}
#| label: fig-subplot-2
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "灵活布局"

p11 <- plotly::subplot(plotly::style(p1, showlegend = FALSE),
  plotly::style(p2, showlegend = FALSE),
  nrows = 1, margin = 0.05, shareY = TRUE, titleX = TRUE
)

plotly::subplot(p11,
  plotly::style(p2, showlegend = FALSE),
  nrows = 2, margin = 0.05, shareY = FALSE, titleX = FALSE
)
```

### 图表联动 {#sec-plotly-crosstalk}

**crosstalk** 包可将 **plotly** 包绘制的图形和 **DT** 包制作的表格联动起来。**plotly** 绘制交互图形，在图形上用套索工具筛选出来的数据显示在表格中。

```{r}
#| label: fig-crosstalk
#| eval: !expr knitr::is_html_output(excludes = 'epub')
#| fig-cap: "图表联动"

library(crosstalk)
# quakes 数据变成可共享的
quakes_sd <- SharedData$new(quakes)
# 绘制交互图形
p <- plotly::plot_ly(quakes_sd, x = ~long, y = ~lat) |> 
  plotly::add_markers() |> 
  plotly::highlight(on = "plotly_selected", off = "plotly_deselect")
# 制作表格
d <- DT::datatable(quakes_sd, options = list(dom = "tp"))
# 将图表组合一起展示
bscols(list(p, d))
```