def test_update_fdist(self):
filtered_words = utils.tokenize_and_filter(self.sr)
fdist = utils.get_freq_dist(filtered_words)
# take distribution and send it empty list
fdist2 = update_fdist(fdist, [])
self.assertEqual(fdist, fdist2)
time.sleep(5)
self.g.latitude = 40.734073
self.g.longitude = -73.990663
self.g.count = 100
self.sr = self.g.search()
filtered_words = utils.tokenize_and_filter(self.sr)
# updating with entirely new word set -> should be longer
old_len_fdist = len(fdist)
fdist = update_fdist(fdist, filtered_words)
self.assertTrue(len(fdist) > old_len_fdist)
def test_update_fdist(self):
filtered_words = utils.tokenize_and_filter(self.sr)
fdist = utils.get_freq_dist(filtered_words)
# take distribution and send it empty list
fdist2 = update_fdist(fdist, [])
self.assertEqual(fdist, fdist2)
time.sleep(5)
self.g.latitude = 40.734073
self.g.longitude = -73.990663
self.g.count = 100
self.sr = self.g.search()
filtered_words = utils.tokenize_and_filter(self.sr)
# updating with entirely new word set -> should be longer
old_len_fdist = len(fdist)
fdist = update_fdist(fdist, filtered_words)
self.assertTrue(len(fdist) > old_len_fdist)
def updating_plot(geosearchclass, number_of_words, grow=True):
search_results = geosearchclass.search()
filtered_words = utils.tokenize_and_filter(search_results)
fdist = utils.get_freq_dist(filtered_words)
# set up plot
samples = [item for item, _ in fdist.most_common(number_of_words)]
freqs = [fdist[sample] for sample in samples]
plt.grid(True, color="silver")
plt.plot(freqs, range(len(freqs)))
plt.yticks(range(len(samples)), [s for s in samples])
plt.ylabel("Samples")
plt.xlabel("Counts")
plt.title("Top Words Frequency Distribution")
plt.ion()
plt.show()
# set up loop
old_ids = set([s.id for s in search_results])
for i in xrange(100):
plt.pause(5)
# use mixed above, change to recent here
geosearchclass.result_type = "recent"
# perturbation study
# if i%2: # for testing purposes
# # #change location every odd time to nyc
# # geosearchclass.latitude =40.734073
# # geosearchclass.longitude =-73.990663
# # perturb latitude
# geosearchclass.latitude =geosearchclass.latitude + .001
# else:
# #now back to sf
# # geosearchclass.latitude = 37.7821
# # geosearchclass.longitude = -122.4093
# geosearchclass.longitude =geosearchclass.longitude + .001
search_results = geosearchclass.search()
new_search_results = utils.new_tweets(search_results, old_ids)
if new_search_results:
filtered_words = utils.tokenize_and_filter(new_search_results)
fdist = update_fdist(fdist, filtered_words)
if grow:
newsamples = [
item for item, _ in fdist.most_common(number_of_words)
]
s1 = set(newsamples)
s2 = set(samples)
s1.difference_update(s2)
if s1:
print "New words: " + str(list(s1))
newsamples = list(s1)
samples.extend(newsamples)
plt.yticks(range(len(samples)), [s for s in samples])
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
if grow:
plt.draw()
print '%d new tweet(s)' % len(new_search_results)
old_ids.update(set([s.id for s in new_search_results]))
else:
print "no updates"
def updating_stream_plot(q, number_of_words=30):
"""This plot uses the streaming API to get real time twitter
information from a given region, determined by a geo-coordinate
bounding box. The upper left and lower right determine the
bounding box.
q is a queue instance, which holds tweets
number_of_words determines the average number of words in the
plot. Once the plot reaches 2 x number_of_words, it is shrunk down
to the new set of words and starts growing again
To exit the program early, hit CTRL + Z to stop the python script
and then CTRL + D twice to kill the terminal process and close the
window.
"""
setup = False
fdist = None
samples = None
draw_time = 0.1
samples = []
plt.ion()
plt.grid(True, color="silver")
for i in range(100000):
status = q.get()
search_results = [status]
while not q.empty():
print "getting another tweet"
status = q.get()
search_results.append(status)
if not setup:
print "Gathering enough data to begin plotting"
while len(samples) < 1:
status = q.get()
search_results.append(status)
filtered_words = utils.tokenize_and_filter(search_results)
if fdist is None:
fdist = utils.get_freq_dist(filtered_words)
else:
fdist = update_fdist(fdist, filtered_words)
n_words = min(10, len(fdist))
samples = [item for item, _ in fdist.most_common(n_words)]
# print "len(samples) = {}".format(len(samples))
samples = remove_infrequent_words(samples, fdist)
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
plt.yticks(range(len(samples)), [s for s in samples])
plt.ylabel("Samples")
plt.xlabel("Counts")
plt.title("Top Words Frequency Distribution")
plt.show()
plt.pause(draw_time)
setup = True
else:
filtered_words = utils.tokenize_and_filter(search_results)
fdist = update_fdist(fdist, filtered_words)
newsamples = [
item for item, _ in fdist.most_common(number_of_words)
]
newsamples = remove_infrequent_words(newsamples, fdist)
s1 = set(newsamples)
s2 = set(samples)
s1.difference_update(s2)
if s1:
print "New words: " + str(list(s1))
newsamples = list(s1)
samples.extend(newsamples)
if len(samples) > 2 * number_of_words:
samples = newsamples
plt.close()
plt.yticks(range(len(samples)), [s for s in samples])
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
plt.draw()
plt.pause(draw_time)
kill_plot()
return
def updating_plot(geosearchclass, number_of_words, grow=True):
search_results = geosearchclass.search()
filtered_words = utils.tokenize_and_filter(search_results)
fdist = utils.get_freq_dist(filtered_words)
# set up plot
samples = [item for item, _ in fdist.most_common(number_of_words)]
freqs = [fdist[sample] for sample in samples]
plt.grid(True, color="silver")
plt.plot(freqs, range(len(freqs)))
plt.yticks(range(len(samples)), [s for s in samples])
plt.ylabel("Samples")
plt.xlabel("Counts")
plt.title("Top Words Frequency Distribution")
plt.ion()
plt.show()
# set up loop
old_ids = set([s.id for s in search_results])
for i in xrange(100):
plt.pause(5)
# use mixed above, change to recent here
geosearchclass.result_type = "recent"
# perturbation study
# if i%2: # for testing purposes
# # #change location every odd time to nyc
# # geosearchclass.latitude =40.734073
# # geosearchclass.longitude =-73.990663
# # perturb latitude
# geosearchclass.latitude =geosearchclass.latitude + .001
# else:
# #now back to sf
# # geosearchclass.latitude = 37.7821
# # geosearchclass.longitude = -122.4093
# geosearchclass.longitude =geosearchclass.longitude + .001
search_results = geosearchclass.search()
new_search_results = utils.new_tweets(search_results, old_ids)
if new_search_results:
filtered_words = utils.tokenize_and_filter(new_search_results)
fdist = update_fdist(fdist, filtered_words)
if grow:
newsamples = [item
for item, _ in fdist.most_common(number_of_words)
]
s1 = set(newsamples)
s2 = set(samples)
s1.difference_update(s2)
if s1:
print "New words: " + str(list(s1))
newsamples = list(s1)
samples.extend(newsamples)
plt.yticks(range(len(samples)), [s for s in samples])
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
if grow:
plt.draw()
print '%d new tweet(s)' % len(new_search_results)
old_ids.update(set([s.id for s in new_search_results]))
else:
print "no updates"
def updating_stream_plot(q, number_of_words=30):
"""This plot uses the streaming API to get real time twitter
information from a given region, determined by a geo-coordinate
bounding box. The upper left and lower right determine the
bounding box.
q is a queue instance, which holds tweets
number_of_words determines the average number of words in the
plot. Once the plot reaches 2 x number_of_words, it is shrunk down
to the new set of words and starts growing again
To exit the program early, hit CTRL + Z to stop the python script
and then CTRL + D twice to kill the terminal process and close the
window.
"""
setup = False
fdist = None
samples = None
draw_time = 0.1
samples = []
plt.ion()
plt.grid(True, color="silver")
for i in range(100000):
status = q.get()
search_results = [status]
while not q.empty():
print "getting another tweet"
status = q.get()
search_results.append(status)
if not setup:
print "Gathering enough data to begin plotting"
while len(samples) < 1:
status = q.get()
search_results.append(status)
filtered_words = utils.tokenize_and_filter(search_results)
if fdist is None:
fdist = utils.get_freq_dist(filtered_words)
else:
fdist = update_fdist(fdist, filtered_words)
n_words = min(10, len(fdist))
samples = [item for item, _ in fdist.most_common(n_words)]
# print "len(samples) = {}".format(len(samples))
samples = remove_infrequent_words(samples, fdist)
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
plt.yticks(range(len(samples)), [s for s in samples])
plt.ylabel("Samples")
plt.xlabel("Counts")
plt.title("Top Words Frequency Distribution")
plt.show()
plt.pause(draw_time)
setup = True
else:
filtered_words = utils.tokenize_and_filter(search_results)
fdist = update_fdist(fdist, filtered_words)
newsamples = [item
for item, _ in fdist.most_common(number_of_words)]
newsamples = remove_infrequent_words(newsamples, fdist)
s1 = set(newsamples)
s2 = set(samples)
s1.difference_update(s2)
if s1:
print "New words: " + str(list(s1))
newsamples = list(s1)
samples.extend(newsamples)
if len(samples) > 2*number_of_words:
samples = newsamples
plt.close()
plt.yticks(range(len(samples)), [s for s in samples])
freqs = [fdist[sample] for sample in samples]
plt.plot(freqs, range(len(freqs)))
plt.draw()
plt.pause(draw_time)
kill_plot()
return