Joshuaek

from sklearn.neural_network import MLPRegressor
from nltk.stem.lancaster import LancasterStemmer
from nltk.corpus import stopwords
import re
from operator import itemgetter

def prepareSentence(s):
    stemmer = LancasterStemmer()
    ignore_words = set(stopwords.words('english'))
    regpattern = re.compile('[\W_]+" "')

Joshuaek

import tweepy
import json
import datetime
import time

# API access details - get yours from Twitter
access_token = "********************"
access_token_secret = "********************"
consumer_key = "********************"
consumer_secret = "********************"

Joshuaek

import glob
import pathlib
import os
import json

#Make sure we're in the same dir that has the downloaded JSON twitter data
os.chdir('/home/username/twitter_data')

tweets = []

Joshuaek

words = []

for tweet in tweets:
    words.extend(prepareSentence(tweet["text"]))

distinct_words = set(words)

lower_threshold = 10
upper_threshold = 350
counts = []

Joshuaek

def toBOW(sentance, words):
    bag = []
    for word in words:
        bag.append(1) if word in sentance else bag.append(0)
    return bag

Joshuaek

inputs = []
outputs = []

for tweet in tweets:
    sentence = prepareSentence(tweet["text"])
    # create our bag of words array
    bag = toBOW(sentence, final_words)
    inputs.append(bag)
    #Calculate a score, 1 if any engagement, 0 if none

Joshuaek

import feedparser

potential_posts = {}
for url in ['https://feeds.feedburner.com/CoinDesk', 'http://newsrss.bbc.co.uk/rss/newsonline_uk_edition/technology/rss.xml','http://i.perezhilton.com/?feed=rss2']:

    feed = feedparser.parse(url)
    for article in feed.entries:
        sentence = prepareSentence(article.title)
        bag = toBOW(sentence, final_words)
        potential_posts[article.title] = nnet.predict([bag])[0]

Joshuaek

Joshuaek

import os
import sys
import argparse
import fnmatch
import logging
import papermill as pm
from datetime import datetime
import time

Joshuaek