R新手求助：Plotly绘制3D曲面图异常及替代方案咨询

Hey there! Let's break down what's going wrong with your Plotly 3D surface plot, and fix it step by step.

Why your current code isn't working #

The key issue here is that plotly::add_surface() expects a 2D wide-format matrix where:

• Rows correspond to unique x values
• Columns correspond to unique y values
• Each cell holds the corresponding z value (forming a regular grid)

But your d3 matrix is a long-format 3-column matrix (z, y, x in rows), so Plotly is treating the row indices as the x-axis, column indices as the y-axis, and the matrix values as z—hence the mismatched axes you're seeing.

Fixing the Plotly code #

First, you need to convert your long-format data into a regular grid. Since your sample data looks like scattered points (not a pre-defined grid), we'll use interpolation to create a smooth grid. Here's how:

# Load required packages
library(plotly)
library(akima) # For interpolation

# Use your original dataframe (no need to convert to matrix first!)
# Interpolate scattered x/y/z into a regular grid
interpolated_grid <- interp(
  x = mydataframe$x, 
  y = mydataframe$y, 
  z = mydataframe$z,
  nx = 50, ny = 50 # Adjust resolution as needed
)

# Now plot the surface with the correct axes
plot_ly(
  x = interpolated_grid$x,
  y = interpolated_grid$y,
  z = interpolated_grid$z
) %>% 
  add_surface() %>%
  layout(scene = list(xaxis = list(title = "x"), yaxis = list(title = "y"), zaxis = list(title = "z")))

If your data already has a regular grid (each x pairs with every y), you can skip interpolation and use tidyr::pivot_wider to reshape directly:

library(tidyr)

# Reshape to wide format
wide_data <- mydataframe %>%
  select(x, y, z) %>%
  pivot_wider(names_from = y, values_from = z)

# Convert to matrix for Plotly
z_matrix <- as.matrix(wide_data[, -1])
rownames(z_matrix) <- wide_data$x

# Plot
plot_ly(
  x = as.numeric(rownames(z_matrix)),
  y = as.numeric(colnames(z_matrix)),
  z = z_matrix
) %>% add_surface()

Alternative 3D surface plotting methods #

If you want other options beyond Plotly, here are two solid choices:

1. Base R's persp() (static plots)

Great for quick static 3D surfaces with customizable color schemes similar to regression plots:

# Use the interpolated grid from earlier
persp(
  interpolated_grid$x,
  interpolated_grid$y,
  interpolated_grid$z,
  col = terrain.colors(100), # Warm/cool gradient palette
  theta = 30, phi = 30, # Adjust view angle
  xlab = "x", ylab = "y", zlab = "z"
)

2. rgl package (interactive 3D plots)

For fully interactive 3D surfaces that you can rotate, zoom, and pan:

library(rgl)

# Create interactive surface
surface3d(
  interpolated_grid$x,
  interpolated_grid$y,
  interpolated_grid$z,
  col = heat.colors(100) # Customizable color gradient
)

# Export as an interactive HTML widget (optional)
rglwidget()

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