Search code examples
pythonmatplotlibstatisticslinear-regression

Plot linear model in 3d with Matplotlib

I'm trying to create a 3d plot of a linear model fit for a data set. I was able to do this relatively easily in R, but I'm really struggling to do the same in Python. Here is what I've done in R:

3d plot

Here's what I've done in Python:

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import statsmodels.formula.api as sm

csv = pd.read_csv('http://www-bcf.usc.edu/~gareth/ISL/Advertising.csv', index_col=0)
model = sm.ols(formula='Sales ~ TV + Radio', data = csv)
fit = model.fit()

fit.summary()

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')

ax.scatter(csv['TV'], csv['Radio'], csv['Sales'], c='r', marker='o')

xx, yy = np.meshgrid(csv['TV'], csv['Radio'])

# Not what I expected :(
# ax.plot_surface(xx, yy, fit.fittedvalues)

ax.set_xlabel('TV')
ax.set_ylabel('Radio')
ax.set_zlabel('Sales')

plt.show()

What am I doing wrong and what should I do instead?

Thank you.

Solution

  • You were correct in assuming that plot_surface wants a meshgrid of coordinates to work with, but predict wants a data structure like the one you fitted with (the "exog"). 

    exog = pd.core.frame.DataFrame({'TV':xx.ravel(),'Radio':yy.ravel()})
out = fit.predict(exog=exog)
ax.plot_surface(xx, yy, out.reshape(xx.shape), color='None')