tsne gist

tsne.py

from sklearn.datasets import fetch_mldata
from sklearn.manifold import TSNE
from sklearn.decomposition import PCA
import seaborn as sns
import numpy as np
import matplotlib.pyplot as plt

# get mnist data
mnist = fetch_mldata("MNIST original")
X = mnist.data / 255.0
y = mnist.target

# first reduce dimensionality before feeding to t-sne
pca = PCA(n_components=50)
X_pca = pca.fit_transform(X)

# randomly sample data to run quickly
rows = np.arange(70000)
np.random.shuffle(rows)
n_select = 10000

# reduce dimensionality with t-sne
tsne = TSNE(n_components=2, verbose=1, perplexity=50, n_iter=1000, learning_rate=200)
tsne_results = tsne.fit_transform(X_pca[rows[:n_select],:])

# visualize
df_tsne = pd.DataFrame(tsne_results, columns=['comp1', 'comp2'])
df_tsne['label'] = y[rows[:n_select]]
sns.lmplot(x='comp1', y='comp2', data=df_tsne, hue='label', fit_reg=False)