for index, learning_decay in enumerate(learning_decays):
(c1, c2) = color_couples[index]
# prepare the scores array (y-axis)
d = dict_learningDecay[str(learning_decay) + '_learning_decay']
train_scores = np.array(d['train_scores'])
test_scores = np.array(d['test_scores'])
train_perplexities = np.array(d['train_perplexities'])
test_perplexities = np.array(d['test_perplexities'])
alpha = 0.1
label = 'learning_decay ' + str(learning_decay)
plot_scores(ax_1,
x_axis,
train_scores,
n_splits,
c1,
title='train_scores',
label=label)
plot_scores(ax_2,
x_axis,
train_perplexities,
n_splits,
c1,
title='train_perplexities',
label=label)
plot_scores(ax_3,
x_axis,
test_scores,
n_splits,
c2,
color_couples = [('#99c9eb', '#f998a5'), ('#4ca1dd', '#f77687'),
('#0079cf', '#f6546a'),
('#005490', '#c44354'), ('#003052', '#93323f'), ]
x_axis = range(0, max_iter, valid_iter)
plt.figure(figsize=(20, 16))
plt.suptitle('Tuning n_topic (learning decay = '+str(learning_decay)+')', fontsize=12)
ax_1 = plt.subplot(221)
ax_2 = plt.subplot(222)
ax_3 = plt.subplot(223)
ax_4 = plt.subplot(224)
for index, n_component in enumerate(n_components):
(c1, c2) = color_couples[index]
d = dict_num_topic[str(n_component) + '_topics']
train_scores = np.array(d['train_scores'])
test_scores = np.array(d['test_scores'])
train_perplexities = np.array(d['train_perplexities'])
test_perplexities = np.array(d['test_perplexities'])
alpha = 0.1
label = 'topic_'+str(n_component)
plot_scores(ax_1, x_axis, train_scores, n_splits, c1, title='train_scores', label=label)
plot_scores(ax_2, x_axis, train_perplexities, n_splits, c1, title='train_perplexities', label=label)
plot_scores(ax_3, x_axis, test_scores, n_splits, c2, title='test_scores', label=label)
plot_scores(ax_4, x_axis, test_perplexities, n_splits, c2, title='test_perplexities', label=label)
plt.savefig('converge_exploration_nTopic(learning decay'+str(learning_decay)+'_full.png')
print "\nFinish Plotting within", time() - start_time, 'secends'