def plot_scores(x_s, y_s, max_x=np.inf, color='g.', label=None, clear=True):
plot_xs = sorted(x_s[x_s < max_x])
indices = sorted(range(len(plot_xs)), cmp=ind_cmp(x_s))
if clear:
clf()
#plot(plot_xs, [tc_scores[i] for i in indices], '.')
#plot(plot_xs, [pc_scores[i] for i in indices], 'r.-')
return plot(plot_xs, [y_s[i] for i in indices], color, label=label)[0]
def which_topics(words, voc_size=1000):
which_topics = {}
sig_words = get_sig_words(words)
top_words = [sorted(range(len(topic)), cmp=ind_cmp(topic), reverse=True)
for topic in words]
for word in range(voc_size):
which_topics[word] = []
for i in range(len(top_words)):
if top_words[i].index(word) < sig_words[i]:
which_topics[word].append(i)
return which_topics
def plot_stats(stats, x, algs=None, flat=False):
if algs == None:
algs = ['lda', 'ldaT', 'ldaC', 'projector', 'Baseline', 'LSI']
colors = 'brgmcyk'
clf()
k = int(stats['k'][0])
for i in range(len(algs)):
color = colors[i]
alg = algs[i]
if flat and alg != 'Baseline':
alg += '-' + str(k)
x_s = stats[x]
scores = stats[alg]
indices = sorted(range(len(x_s)), cmp=ind_cmp(x_s))
plot(sorted(x_s), [scores[i] for i in indices], color + '.', label=alg)
legend(loc="best")
def plot_extract(flat_results, n_s=[1000, 1500, 2000, 2500], y='projector-20',
clear=True):
if clear:
clf()
colors = {1000:'b', 1500:'g', 2000:'r', 2500:'k'}
markers = {1000:'+', 1500:'x', 2000:'o', 2500:'.'}
plots = []
for n in n_s:
results = extract(flat_results, n)
indices = sorted(range(len(results['l'])), cmp=ind_cmp(results['l']))
if y == 'ratio':
y_s = np.array([results['projector-20'][i] for i in indices]) / \
np.array([results['ldaT-20'][i] for i in indices])
else:
y_s = [results[y][i] for i in indices]
plots.append(plot(sorted(results['l']), y_s,
colors[n] + '-' + markers[n],
label=str(n) + ' documents')[0])
legend(loc='best')
return plots
def plot_deep_stats(stats, k=20, index=0, clear=True):
"""plot the output of get_deep_stats()
"""
if clear:
clf()
#algs = ['lda', 'ldaT', 'ldaC', 'projector', 'Baseline', 'LSI']
algs = ['lda', 'ldaC', 'projector', 'Baseline', 'LSI', 'kmeans',
'malletlda']
markers = ['.', 'x', 'd', 'o', '-', '*', '^']
colors = 'bgrymkc'
for alg in algs:
if alg != 'Baseline':
alg_name = alg + '-' + str(k)
else:
alg_name = alg
sig_words = sorted(stats[index]['sig_words'])
indices = sorted(range(len(sig_words)),
cmp=ind_cmp(stats[index]['sig_words']))
scores = stats[index][alg_name]
plot(sig_words, [scores[i] for i in indices],
colors[algs.index(alg)] + markers[algs.index(alg)] + '-',
label=alg)
legend(loc='best')
