I was at an interesting talk about sentiment analysis and decided to try something simple myself. Sentiment analysis is a text analytics method that compares text data with a list of words with positive or negative sentiments. The relative frequency of the positive or negative words is a crude measure of the general sentiment of the text item. I ran a sentiment analysis on the text of the famous Charles Dickens novel, A Christmas Carol.

You can get the code for my sentiment analysis on my github site.

Running a tool like sentiment analysis is always difficult, but especially for a work of literature. The rule for good writers, I hear, is that you are better of showing something than saying something. So you get a sense of dread watching Scrooge’s possessions getting sold off that is much stronger than the impact of individual words. Even with this limitation, sentiment analysis can reveal some interesting patterns.

Here’s a graph of the analysis.

The individual dots represent a sentiment value for the individual paragraphs. Notice how you see a mix throughout the book. Every chapter has a large number of paragraphs with positive sentiment values and a large number of paragraphs with negative sentiment values. The blue line, a smooth curve, shows that the balance of positive and negative paragraphs swings back and forth.

The first chapter, for example, starts out slightly positive. This may represent Scrooge’s smug satisfaction with himself. The text turns negative about halfway through. This is when Scrooge goes to his home, encounters a strange vision on his door knocker, and then meets the ghost of Jacob Marley.

The next chapter, representing Scrooge’s encounter with the Spirit of Christmas Past, is a fairly even mix. Recall that Scrooge is encountering memories of his early years, and there is a mix of nostalgia and regret throughout.

The third chapter, representing Scrooge’s encounter with the Spirit of Christmas Present, starts out positively, with Scrooge seeing the happiness in the Cratchet family and at his nephew’s party. But the mood turns very dark, especially at the end of the chapter when Scrooge meets the two children, ignorance and want, hiding under the spirit’s cloak.

The fourth chapter should be the one with the most negative sentiment, but perhaps the imagery and symbolism is not captured sufficiently by the mere words on the page.

The last chapter, where Scrooge awakens and realizes that he has been given a fresh chance at life, is clearly filled with positive sentiments.

If I get a chance, I want to experiment with an interactive display which allows you to hover over a point and see the full paragraph of text associated with that point.

David Robinson. Trump’s Android and iPhone tweets, one year later. Variance Explained blog, August 9, 2017. Available at http://varianceexplained.org/r/trump-followup/.