kernel pca scikit plotting code extension for multiple gammas

[teamiceberg]

@rasbt

I extended the sci-kit kernel PCA for transforming the half moon example to produce plots for multiple gammas. This sequence of plots can then demonstrate how a gamma value of 15 visually makes sense. Thought your readers might find it useful.

Here's the small extension code.

`#Kernel principal component analysis in scikit-learn analyzed for a stream of gammas gamma = [1, 3, 5, 7, 10, 12, 15, 20] X, y = make_moons(n_samples=100, random_state=123) ncols = 2 nrows_tup=divmod(len(gamma),ncols) fig, ax = plt.subplots(nrows=nrows_tup[0] + nrows_tup[1], ncols=ncols) textpos = (0.0, 0.3) for g in enumerate(gamma): scikit_kpca = KernelPCA(n_components=2, kernel='rbf', gamma=g[1]) X_skernpca = scikit_kpca.fit_transform(X) idx = divmod(g[0] , ncols) ax[idx].scatter(X_skernpca[y == 0, 0], X_skernpca[y == 0, 1], color='red', marker='^', alpha=0.5) ax[idx].scatter(X_skernpca[y == 1, 0], X_skernpca[y == 1, 1], color='blue', marker='o', alpha=0.5)

ax[idx].set_xlabel('PC1', fontsize = 8)
ax[idx].set_ylabel('PC2', fontsize = 8)
ax[idx].text(textpos[0], textpos[1], "gamma:"+str(g[1]), fontsize = 6 )

plt.suptitle('Sci-kit RBF kernel PCA: with gamma = ' + str(gamma), va = "bottom", fontsize=8)

plt.tight_layout() plt.show()`

Here's the matplot lib output: