在R 4.2.1中用Plotly 4.10.0绘制多3D散点子图堆叠问题求助

我使用R 4.2.1及Plotly 4.10.0，想要将不同聚类算法的结果制作成6幅3D散点图组成的子图：

• 数据包含V1、V2、V3三个特征列，对应不同聚类标签（取值为1和2）
• 单独绘制单幅3D散点图时，能正常按标签用Set2色系区分标记点
• 但使用subplot()整合6幅图时，所有图形堆叠到了同一幅图中，无法形成独立子图

我的代码如下：

### Atom true labels
atom_true_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                            color=~factor(atom_labels), colors="Set2") %>%
                            add_markers(size=2)

### Atom km labels
atom_km_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(atom_km$cluster), colors="Set2") %>%
                          add_markers(size=2) 

### Atom cl labels
atom_cl_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(atom_cl$clustering), colors="Set2") %>%
                          add_markers(size=2)

### Atom ag_avr labels
atom_ag_avr_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                              color=~factor(cutree(atom_ag_avr, k=groups_num)), colors="Set2") %>%
                              add_markers(size=2)

### Atom ag_wrd labels
atom_ag_wrd_labels <- plot_ly(atom, x=~V1, y=~V2,z=~V3,
                              color=~factor(cutree(atom_ag_wrd, k=groups_num)), colors="Set2") %>%
                              add_markers(size=2)

### Atom di labels
atom_di_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(cutree(atom_di, k=groups_num)), colors="Set2") %>%
                          add_markers(size=2)

fig <- subplot(atom_true_labels, atom_km_labels, atom_cl_labels, atom_ag_avr_labels, atom_ag_wrd_labels, atom_di_labels)
fig %>% layout(title="Atom data clustering")

堆叠问题的核心原因是Plotly处理3D图时，默认共享坐标系，导致子图重叠。通过为每个子图配置独立坐标系、明确子图布局结构即可解决：

修改后的代码如下：

# 定义统一的3D坐标系配置，确保每个子图坐标轴一致
scene_config <- list(
  xaxis = list(title = "V1"),
  yaxis = list(title = "V2"),
  zaxis = list(title = "V3")
)

### Atom true labels
atom_true_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                            color=~factor(atom_labels), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "True Labels")

### Atom km labels
atom_km_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(atom_km$cluster), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "K-Means")

### Atom cl labels
atom_cl_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(atom_cl$clustering), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "CLARA")

### Atom ag_avr labels
atom_ag_avr_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                              color=~factor(cutree(atom_ag_avr, k=groups_num)), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "Agglomerative (Average)")

### Atom ag_wrd labels
atom_ag_wrd_labels <- plot_ly(atom, x=~V1, y=~V2,z=~V3,
                              color=~factor(cutree(atom_ag_wrd, k=groups_num)), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "Agglomerative (Ward)")

### Atom di labels
atom_di_labels <- plot_ly(atom, x=~V1, y=~V2, z=~V3,
                          color=~factor(cutree(atom_di, k=groups_num)), colors="Set2") %>%
  add_markers(size=2) %>%
  layout(scene = scene_config, title = "Diana")

# 明确子图排列结构，关闭坐标轴共享
fig <- subplot(
  atom_true_labels, atom_km_labels, atom_cl_labels, 
  atom_ag_avr_labels, atom_ag_wrd_labels, atom_di_labels,
  rows = 2, cols = 3,
  shareX = FALSE, shareY = FALSE,
  titleX = TRUE, titleY = TRUE
)

# 调整整体布局标题和间距
fig %>% layout(
  title = list(text = "Atom data clustering", x = 0.5),
  margin = list(l = 50, r = 50, b = 50, t = 80)
)

• 为每个子图添加独立的scene配置，避免共享3D坐标系
• 在subplot()中指定rows和cols，明确子图的排列方式（如2行3列）
• 设置shareX=FALSE和shareY=FALSE，确保每个子图坐标轴独立
• 为每个子图添加单独标题，方便区分不同聚类算法结果

内容的提问来源于stack exchange，提问作者Kamil Niewulis