How to plot an eigenspectrum in python?

Question

This is what I have tried:

# Show the eigenspectrum
eigenvalues = pca.explained_variance_
print("The eigenvalues:\n\t", eigenvalues)

idx = eigenvalues.argsort()
print(idx)

plt.plot(idx, color='green', marker='D')
plt.ylabel('Eigenspectrum')
plt.show()

The shape of result is (640, 2), but what I keep getting is just a straight line.

Could someone please help?

Just to add, I ran a PCA on the data, and plotted a scatter plot of the data successfully. I am not sure how to extract all the eigenvalues, sort them and put into an eigenspectrum.

pca=PCA(n_components=2)
pca.fit(keytrain[:,0:-1])
keytrain_T=pca.transform(keytrain[:,0:-1])

print("Shape of result:", keytrain_T.shape)

# plot the results along with the labels
fig, ax = plt.subplots()
im = ax.scatter(keytrain_T[:, 0], keytrain_T[:, 1], c=y)
fig.colorbar(im);
plt.show()

if `idx` has only two values, why would you expect to see more than 2 points (which is a straight line) when you plot it? — Julien Spronck, Apr 02 '15 at 21:56
I have run a PCA on this, and made a scatter plot of the data with more values. — user4476006, Apr 02 '15 at 21:58
I have added the code and more explanation in the post. To be honest, the eigenvalues are confusing me. — user4476006, Apr 02 '15 at 22:00
That's a different question ... don't think I can help much ... sorry — Julien Spronck, Apr 02 '15 at 22:04
You've set n_components=2, so you get 2 components, and an eigenvalue for the explained variance of each component. — KarlM, Apr 02 '15 at 23:30

score -3 · Answer 1 · edited Jan 18 '17 at 01:06

-3

To get all eigenvalues for a rectangular array, use:

numpy.linalg.eigvals

and for a square matrix use

numpy.linalg.eig

edited Jan 18 '17 at 01:06

zx485

28,498
28
50
59

answered Jan 18 '17 at 00:47

AnandJ

346
4
15

How to plot an eigenspectrum in python?

1 Answers1