How does the performance of various supervised ML and rule-based methods compare when tasked with identifying sentiment in Tweets to U.S. airlines?
Read my blog post for a full discussion, or check out the slides from my Metis presentation.
The table below provides high-level overviews of what each analysis script does. More information (including specific input/ouput data) can be found in each script's header.
| Program | Description |
|---|---|
| 01-Apply-Sentiment.py | Add rule-based TextBlob and VADER sentiment polarity scores to CrowdFlower data. |
| 02-Text-Blob-Sentiment-Analysis.ipynb | Sentiment analysis of 1,000 tweets to United Airlines using TextBlob. |
| 03-Airline-Sentiment-Analysis.ipynb | Compare performance of rule-based and supervised ML-based sentiment classifiers. |
Before running any of these scripts, please make sure to have downloaded:
NumPy, Pandas, Sci-Kit Learn, Matplotlib, Seaborn, Pylab, re, NLTK (including their twitter samples corpus), TextBlob, vaderSentiment, Pickle, Cnfg, Os, TweePy (Twitter API Python Wrapper), MongoDB and PyMongo