plotly 3D Topographical 3D Surface Plot with pandas dataframe

Question

I have a coordenates points cloud in a padas dataframe

Following, I want to plot a ploty surface diagram but Surface method needs a 2D array in z param and i don't know how to build it.

go.Figure(data=[go.Surface(z=z, x=x, y=y)])

Answer 1

I think you need to grid your data first. I created an example from your data.

import pandas as pd
import numpy as np
import plotly.graph_objects as go 
from scipy.interpolate import griddata

#Read data
data = pd.read_csv('data.txt', header=0, delimiter='\t')

#Create meshgrid for x,y
xi = np.linspace(min(data['x']), max(data['x']), num=100)
yi = np.linspace(min(data['y']), max(data['y']), num=100)

x_grid, y_grid = np.meshgrid(xi,yi)

#Grid data
z_grid = griddata((data['x'],data['y']),data['z'],(x_grid,y_grid),method='cubic')

# Plotly 3D Surface
fig = go.Figure(go.Surface(x=x_grid,y=y_grid,z=z_grid,
                           colorscale='viridis',showlegend=True)
)

fig.show()

3D surface from your example data

You can use different methods for gridding (cubic, linear, nearest) https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.griddata.html

Answer 2

It's definitely a bad design that go.Surface doesn't support directly using DataFrames as parameters. Imagine you swapped x and y, for example: x=df.index, y=df.columns, z=df.values. You won't notice that your x-axis data isn't fully displayed if you don't scrutinize it carefully. That's because you reversed x and y, which gives you the wrong result!