An interesting use of the “bing” lexicon and R web scraping would be to look at the minutes of the Federal Open Market Committee as they discuss the economy and the likely path of interest rates going forward.
https://www.federalreserve.gov/monetarypolicy/fomccalendars.htm
I’ve truncated this run as of year end 2016. As you can see the output is highly variable of late, and much of the commentary through late 2015 and 2016 was less bullish than in prior years. Indeed, the regression line suggests a general downward trend in tone since 2012.
I think a better implementation of this approach (besides spending more time with ggplot2 building a more readable chart) would be to use a market-driven lexicon, vs the one I’m using here.
Many thanks to the makers of the tidytext libray who made this possible:
https://cran.r-project.org/web/packages/tidytext/vignettes/tidytext.html
Investing with Data Science 