VADER Sentiment Analysis
Valence Aware Dictionary and sEntiment Reasoner
VADER Sentiment Analysis: Slide Deck
How do I get VADER?
Simple command line install:
> pip install vaderSentiment
If you already have VADER, you can upgrade to the latest version:
> pip install --upgrade vaderSentiment
What do I get with VADER?
VADER package includes:
Comprehensive sentiment lexicon
Python code for rule-based sentiment analysis engine
Files with text snippets and human sentiment ratings from Twitter, New York Times, movie reviews, and Amazon product reviews.
What kind of data can VADER analyze?
Micro-blog length sentences
- Tweets
- Facebook Comments
- News Comments
- Blog Comments
What can VADER tell me about my data?
"VADER gives you a compound score of whether a sentence leans in the direction of positive, neutral, or negative."
How?
- Sums the valence scores of each word in the lexicon, adjusts according to the rules (all those files), and then normalizes to be between -1 and +1
- VADER calculates a compound score using ratios from the portions of text in each category.
Positive >= 0.05Neutral between -0.05 and 0.05Negative <= -0.05
How does VADER actually work?
- VADER analyzes a column of text in a CSV and assigns scores to each text snippet.
For example:
- Comment regarding Antonin Scalia death: “Good riddance! He was the worst thing that ever happened to the Supreme Court. He was against everything including marriage equality, Obamacare, affirmative action, immigration reform etc... Now Obama gets to change the court for a very long time.” VADER analysis: {'neg': 0.089, 'neu': 0.763, 'pos': 0.148, 'compound': 0.4003}
| Pros | Cons |
|---|---|
| *Open Source | *Not accurate for complex data |
| *On Github | *Does not recognize context |
| *Easy to acquire and replicate | *Requires additional tools for visualizing output (pandas, etc.) |
| *Among the most comprehensive tools for social media analysis | |
| You COULD train it further with time and skills |