Code for building a binary classification model in Python to classify Tweets as either related to individual consumer desire or not.

• Jupyter notebooks
  • 01_parse_tweets.ipynb
  • 02_model_intent.ipynb: build and pickle classification model
• labels_tesla.txt: key of labels used in '...labeled.csv' below
• tesla_tweets_rand_for_label_2017-1-6_labeled.csv: labled tweets
• raw tweets in JSON line format
  • teslatweet_2017-1-5.gz
  • teslatweet_2017-1-6.gz
• Pickled models
  • twitter-consumer-desire-logreg.pkl
  • twitter-consumer-desire-pipe.pkl: use this one; it includes feature generation

Symbols used in the label column (in label: definition format)

• 0: can't judge (in another language, only a link, etc)
• n: not about tesla company/product
• m: only about Musk himself
• v: about business, news, current events
• i: by an individual
• c: by a company
• d: individual consumer desire/opinion
• b: intent to buy
• + -: positive or negative sentiment