def random_word_tuple(input_string):
input_list = cleanup.clean_text(input_string)
tuples_histogram = histogram.tuples_histogram(input_list)
index = random.randrange(len(input_list))
value = 0
hist_ind = 0
while value <= index and hist_ind < len(tuples_histogram):
value += tuples_histogram[hist_ind][1]
hist_ind += 1
return tuples_histogram[hist_ind - 1][0]
def load_sentqs_tweets():
if os.path.isfile("data/sentqs_preprocessed.npz"):
loaded_data = np.load("data/sentqs_preprocessed.npz")
return loaded_data['cleaned_tweets'],loaded_data["tweets"], loaded_data['y'],loaded_data['sentiment'],loaded_data["source_idx"],loaded_data["target_idx"]
else:
hashtags = ['ADBE', 'GOOGL', 'AMZN', 'AAPL', 'ADSK', 'BKNG', 'EXPE', 'INTC', 'MSFT', 'NFLX', 'NVDA', 'PYPL', 'SBUX',
'TSLA', 'XEL', 'positive', 'bad', 'sad']
# Loading and preprocessing of tweets
df = pd.read_csv("Tweets.csv")
sentiment = pd.to_numeric(df.iloc[:, -1], errors="raise", downcast="float")
labels,tweets,sentiment = seperate_tweets(df.iloc[:, 1],hashtags,sentiment)
cleaned_tweets = cleanup.clean_text(tweets)
y = preprocessing.LabelEncoder().fit_transform(labels)
source_idx,target_idx = create_domain_adaptation_index(tweets,labels,sentiment)
np.savez_compressed("data/sentqs_preprocessed.npz",tweets=tweets, cleaned_tweets=cleaned_tweets, y=y,sentiment=sentiment,source_idx=source_idx,target_idx=target_idx)
return cleaned_tweets,tweets, y,sentiment,source_idx,target_idx
def main_preprocessing():
hashtags = [
'ADBE', 'GOOGL', 'AMZN', 'AAPL', 'ADSK', 'BKNG', 'EXPE', 'INTC',
'MSFT', 'NFLX', 'NVDA', 'PYPL', 'SBUX', 'TSLA', 'XEL'
]
# Loading and preprocessing of tweets
df = pd.read_csv("Tweets.csv")
labels, tweets = seperate_tweets(df.iloc[:, 1], hashtags)
cleaned_tweets = cleanup.clean_text(tweets)
y = preprocessing.LabelEncoder().fit_transform(labels)
#
# # Get some statistics of the dataset
describe_dataset(cleaned_tweets, labels)
#
# # Create feature representation: TFIDF Variants and skipgram embedding with 1000 dimension and negative sampling
create_representation(cleaned_tweets, y)
# Plot eigenspectrum of embeddings
X = np.load("data/nsdqs_skipgram_embedding.npy")
plot_eigenspectrum(X)
# Plot representation of 2 dimensional tsne embedding
plot_tsne(X, labels)
unique_list.append(item)
for item in unique_list:
occurance = iterable.count(item)
dictionary[item] = [str(item), occurance]
return dictionary
def frequency(histogram_in, item):
'''
This function takes a item and histogram argument and
returns the number of times that item appears in a text.
'''
if item in histogram_in:
return histogram_in[item][1]
else:
return 0
def all_items(histogram_in):
'''Calculates the total number of items in the text.'''
total = 0
for item in histogram_in:
total += histogram_in[item][1]
return total
if __name__ == "__main__":
cleaned_text = clean_text('the_book.txt')
print(histogram(cleaned_text))
from flask import Flask, render_template, redirect, request, url_for
from cleanup import read_in_data, clean_text, add_stop, create_sentence
from random import randint, choice
from markov_chain import higher_markov, random_walk
app = Flask(__name__)
source_text = 'static/tumblr_wordz.txt'
words = read_in_data(source_text)
cleaned_text = clean_text(words)
stripped_words = add_stop(cleaned_text)
markov = higher_markov(stripped_words)
@app.route('/')
def index():
# Check for user input, otherwise, print sentence w/ 6 words
usr_input = request.args.get('num')
num_of_words = int(usr_input) if (usr_input != None
and usr_input != "") else 6
# Create sentence with specified num_of_words
output = random_walk(markov, num_of_words)
sentence = create_sentence(output)
return render_template('index.html', sentence=sentence)
if __name__ == '__main__':
def main():
corpus = clean_text('book_1.txt')
print(random_walk_first(corpus))
cume = 0
for word in histogram_in:
cume += histogram_in[word]
if (cume > value):
return word
def sample_list_O_stuff(histogram_in): # stuff means tuples or lists
cap = 0
i = 0
while i < len(histogram_in):
cap += histogram_in[i][1]
i += 1
value = random.randint(0, cap)
cume = 0
index = 0
while index < len(histogram_in):
cume += histogram_in[index][1]
if cume > value:
return histogram_in[index][0]
index += 1
if __name__ == "__main__":
text = clean_text('book_1.txt')
hist = Dictogram(text)
list_hist = Listogram(text)
print(sample_list_O_stuff(list_hist))
print(dictionary_sample(hist))
def __init__(self):
self.text = cleanup.clean_text()
self.token = words_list.token(self.text)
self.word_dict = {}
import cleanup
def token(text):
""" Input: string that is clean text
Process: converts string to list of words
Output: List of words
"""
word_list = list(text.split())
return word_list
if __name__ == "__main__":
text = cleanup.clean_text()
token(text)
word_list = token(text)
# print(word_list)
