06 — Text as Data

Text mining, sentiment analysis, and dictionary methods.

Word.

This module covers text analysis: preprocessing, tokenization, tf-idf, dictionary methods, and applications including the Economic Policy Uncertainty Index.

Lecture slides

View fullscreen | Source

Code

Check the course repository for the R scripts.

Key packages: quanteda, tidytext, stringr

Further resources

• Gentzkow, Kelly, and Taddy (2019): Text as Data (JEL)
• Baker, Bloom, and Davis (2016): Measuring Economic Policy Uncertainty (QJE)
• Julia Silge and David Robinson’s Text Mining with R (free online book)
• quanteda tutorials