def fetch_sotu():
baseurl = 'http://stateoftheunion.onetwothree.net/texts/'
path = 'data/sotu/sotus.pkl'
def download_text(datestr):
pagetext = urlopen(baseurl + datestr + '.html').read().replace('\n', ' ')
paragraphs = re.findall(r'<p>(.*?)</p>', pagetext, re.DOTALL)
return ' '.join(paragraph.strip() for paragraph in paragraphs)
if not os.path.isfile(path):
response = urlopen(baseurl + 'index.html')
dates = re.findall(r'<li><a href="([0-9]+)\.html">', response.read())
print('Downloading SOTU data...')
sotus = {date:download_text(date) for date in progprint(dates)}
print('...done!')
mkdir(os.path.dirname(path))
with open(path, 'w') as outfile:
pickle.dump(sotus, outfile, protocol=-1)
else:
with open(path, 'r') as infile:
sotus = pickle.load(infile)
return sotus
def fetch_sotu():
baseurl = 'http://stateoftheunion.onetwothree.net/texts/'
path = 'data/sotu/sotus.pkl'
def download_text(datestr):
pagetext = urlopen(baseurl + datestr + '.html').read().replace(
'\n', ' ')
paragraphs = re.findall(r'<p>(.*?)</p>', pagetext, re.DOTALL)
return ' '.join(paragraph.strip() for paragraph in paragraphs)
if not os.path.isfile(path):
response = urlopen(baseurl + 'index.html')
dates = re.findall(r'<li><a href="([0-9]+)\.html">', response.read())
print('Downloading SOTU data...')
sotus = {date: download_text(date) for date in progprint(dates)}
print('...done!')
mkdir(os.path.dirname(path))
with open(path, 'w') as outfile:
pickle.dump(sotus, outfile, protocol=-1)
else:
with open(path, 'r') as infile:
sotus = pickle.load(infile)
return sotus
def fetch_sotu():
baseurl = "http://stateoftheunion.onetwothree.net/texts/"
path = "data/sotu/sotus.pkl"
def download_text(datestr):
pagetext = urlopen(baseurl + datestr + ".html").read().replace("\n", " ")
paragraphs = re.findall(r"<p>(.*?)</p>", pagetext, re.DOTALL)
return " ".join(paragraph.strip() for paragraph in paragraphs)
if not os.path.isfile(path):
response = urlopen(baseurl + "index.html")
dates = re.findall(r'<li><a href="([0-9]+)\.html">', response.read())
print "Downloading SOTU data..."
sotus = {date: download_text(date) for date in progprint(dates)}
print "...done!"
mkdir(os.path.dirname(path))
with open(path, "w") as outfile:
pickle.dump(sotus, outfile, protocol=-1)
else:
with open(path, "r") as infile:
sotus = pickle.load(infile)
return sotus
def fit_ldac_ctm(num_topics, datadir, resultsdir):
if isdir(resultsdir):
for f in glob(join(resultsdir, '*.dat')):
os.remove(f)
mkdir(resultsdir)
with open(logfile, 'w') as log:
subprocess.check_call(
[ctm_binary_path, 'est', join(datadir, documentfile),
str(num_topics), 'rand', resultsdir, settingsfile],
stdout=log, stderr=subprocess.STDOUT)
def fit_ldac_ctm(num_topics, datadir, resultsdir):
if isdir(resultsdir):
for f in glob(join(resultsdir, '*.dat')):
os.remove(f)
mkdir(resultsdir)
with open(logfile, 'w') as log:
subprocess.check_call([
ctm_binary_path, 'est',
join(datadir, documentfile),
str(num_topics), 'rand', resultsdir, settingsfile
],
stdout=log,
stderr=subprocess.STDOUT)
def dump_ldac_dataset(train_data, datadir):
mkdir(datadir)
def pairs(row):
return ['{}:{}'.format(
wordid,int(row[0,wordid])) for wordid in row.nonzero()[1]]
def line(row):
return '{} {}\n'.format(row.nnz, ' '.join(pairs(row)))
with open(join(datadir, documentfile), 'w') as outfile:
outfile.writelines(line(row) for row in train_data if row.nnz > 0)
return np.array([row.nnz > 0 for row in train_data])
def dump_ldac_dataset(train_data, datadir):
mkdir(datadir)
def pairs(row):
return [
'{}:{}'.format(wordid, int(row[0, wordid]))
for wordid in row.nonzero()[1]
]
def line(row):
return '{} {}\n'.format(row.nnz, ' '.join(pairs(row)))
with open(join(datadir, documentfile), 'w') as outfile:
outfile.writelines(line(row) for row in train_data if row.nnz > 0)
return np.array([row.nnz > 0 for row in train_data])
def cached(func):
mkdir(cachedir)
cachebase = os.path.join(cachedir, func.__module__ + func.__name__)
def replace_arrays(v):
if isinstance(v, np.ndarray):
return hashlib.sha1(v).hexdigest()
if isinstance(v, scipy.sparse.csr.csr_matrix):
out = hashlib.sha1(v.data)
out.update(v.indices)
out.update(v.indptr)
return out.hexdigest()
return v
@wraps(func)
def wrapped(*args, **kwargs):
argdict = \
{k:replace_arrays(v) for k,v in
inspect.getcallargs(func,*args,**kwargs).iteritems()}
closurevals = \
[replace_arrays(cell.cell_contents) for cell in func.__closure__ or []]
key = str(hash(frozenset(argdict.items() + closurevals)))
cachefile = cachebase + '.' + key
if os.path.isfile(cachefile):
with gzip.open(cachefile, 'r') as infile:
value = pickle.load(infile)
return value
else:
value = func(*args,**kwargs)
with gzip.open(cachefile, 'w') as outfile:
pickle.dump(value, outfile, protocol=-1)
return value
return wrapped
def cached(func):
mkdir(cachedir)
cachebase = os.path.join(cachedir, func.__module__ + func.__name__)
def replace_arrays(v):
if isinstance(v, np.ndarray):
return hashlib.sha1(v).hexdigest()
if isinstance(v, scipy.sparse.csr.csr_matrix):
out = hashlib.sha1(v.data)
out.update(v.indices)
out.update(v.indptr)
return out.hexdigest()
return v
@wraps(func)
def wrapped(*args, **kwargs):
argdict = \
{k:replace_arrays(v) for k,v in
inspect.getcallargs(func,*args,**kwargs).iteritems()}
closurevals = \
[replace_arrays(cell.cell_contents) for cell in func.__closure__ or []]
key = str(hash(frozenset(argdict.items() + closurevals)))
cachefile = cachebase + '.' + key
if os.path.isfile(cachefile):
with gzip.open(cachefile, 'r') as infile:
value = pickle.load(infile)
return value
else:
value = func(*args,**kwargs)
with gzip.open(cachefile, 'w') as outfile:
pickle.dump(value, outfile, protocol=-1)
return value
return wrapped
"var max iter 20",
"cg max iter -1",
"em convergence 1e-3",
"var convergence 1e-6",
"cg convergence 1e-6",
"lag 1",
"covariance estimate mle",
]
if not has_ctm_c:
msg = 'Please download ctm-c from {url} to {ctmdir} and build it. ' \
'(i.e. the ctm binary should be at {ctm_binary_path})'.format(
url=ctm_url, ctmdir=ctmdir, ctm_binary_path=ctm_binary_path)
raise Exception('\n' + '\n'.join(wrap(msg, 82)))
mkdir(os.path.dirname(settingsfile))
with open(settingsfile, 'w') as outfile:
outfile.writelines('\n'.join(settings))
def dump_ldac_dataset(train_data, datadir):
mkdir(datadir)
def pairs(row):
return ['{}:{}'.format(
wordid,int(row[0,wordid])) for wordid in row.nonzero()[1]]
def line(row):
return '{} {}\n'.format(row.nnz, ' '.join(pairs(row)))
with open(join(datadir, documentfile), 'w') as outfile:
"var max iter 20",
"cg max iter -1",
"em convergence 1e-3",
"var convergence 1e-6",
"cg convergence 1e-6",
"lag 1",
"covariance estimate mle",
]
if not has_ctm_c:
msg = 'Please download ctm-c from {url} to {ctmdir} and build it. ' \
'(i.e. the ctm binary should be at {ctm_binary_path})'.format(
url=ctm_url, ctmdir=ctmdir, ctm_binary_path=ctm_binary_path)
raise Exception('\n' + '\n'.join(wrap(msg, 82)))
mkdir(os.path.dirname(settingsfile))
with open(settingsfile, 'w') as outfile:
outfile.writelines('\n'.join(settings))
def dump_ldac_dataset(train_data, datadir):
mkdir(datadir)
def pairs(row):
return [
'{}:{}'.format(wordid, int(row[0, wordid]))
for wordid in row.nonzero()[1]
]
def line(row):
return '{} {}\n'.format(row.nnz, ' '.join(pairs(row)))
print("Singular vector ", d, " Singular value, ", S[d])
print("Right: ")
print(top_k(5, pi_vd))
print("Left: ")
print(top_k(5, pi_ud))
if __name__ == "__main__":
run = 1
results_dir = os.path.join("results", "alice", "run%03d" % run)
# Make sure the results directory exists
from pgmult.utils import mkdir
if not os.path.exists(results_dir):
print("Making results directory: ", results_dir)
mkdir(results_dir)
# Load the data
Xs, words = load()
# N_docs = 1
docs = slice(0,1)
T_end = 4000
T_split = 100
# Filter out documents shorter than 2 * T_split
Xfilt = [X for X in Xs if X.shape[0] > T_end]
Xtrain = [X[:T_end-T_split] for X in Xfilt[docs]]
Xtest = [X[T_end-T_split:T_end] for X in Xfilt[docs]]
# Perform inference for a range of latent state dimensions and models
legendargs={
"columnspacing": 0.75,
"handletextpad": 0.1
})
fig.savefig(os.path.join(results_dir, "legend.pdf"))
if __name__ == "__main__":
run = 5
results_dir = os.path.join("results", "dna", "run%03d" % run)
# Make sure the results directory exists
from pgmult.utils import mkdir
if not os.path.exists(results_dir):
print("Making results directory: ", results_dir)
mkdir(results_dir)
# Load data
Xs, key = load_data()
# Split data into two
T_end = Xs[0].shape[0]
T_split = 10
Xtrain = [Xs[0][:T_end - T_split, :]]
Xtest = Xs[0][T_end - T_split:T_end, :]
K = len(key)
# Perform inference for a range of latent state dimensions and models
N_samples = 1000
all_results = []
# Ds = np.array([2, 3, 4, 5, 6])
