Loved by learners at thousands of companies
From social media to product reviews, text is an increasingly important type of data across applications, including marketing analytics. In many instances, text is replacing other forms of unstructured data due to how inexpensive and current it is. However, to take advantage of everything that text has to offer, you need to know how to think about, clean, summarize, and model text. In this course, you will use the latest tidy tools to quickly and easily get started with text. You will learn how to wrangle and visualize text, perform sentiment analysis, and run and interpret topic models.
Since text is unstructured data, a certain amount of wrangling is required to get it into a form where you can analyze it. In this chapter, you will learn how to add structure to text by tokenizing, cleaning, and treating text as categorical data.
While counts are nice, visualizations are better. In this chapter, you will learn how to apply what you know from ggplot2 to tidy text data.Plotting word counts50 xpVisualizing complaints100 xpVisualizing non-complaints100 xpImproving word count plots50 xpAdding custom stop words100 xpVisualizing word counts using factors100 xpFaceting word count plots50 xpCounting by product and reordering100 xpVisualizing word counts with facets100 xpPlotting word clouds50 xpCreating a word cloud100 xpAdding a splash of color100 xp
While word counts and visualizations suggest something about the content, we can do more. In this chapter, we move beyond word counts alone to analyze the sentiment or emotional valence of text.Sentiment dictionaries50 xpCounting the NRC sentiments100 xpVisualizing the NRC sentiments100 xpAppending dictionaries50 xpCounting sentiment100 xpVisualizing sentiment100 xpImproving sentiment analysis50 xpPracticing reshaping data100 xpPracticing with grouped summaries100 xpVisualizing sentiment by complaint type100 xp
In this final chapter, we move beyond word counts to uncover the underlying topics in a collection of documents. We will use a standard topic model known as latent Dirichlet allocation.Latent Dirichlet allocation50 xpTopics as word probabilities100 xpSummarizing topics100 xpVisualizing topics100 xpDocument term matrices50 xpCreating a DTM100 xpEvaluating a DTM as a matrix100 xpRunning topic models50 xpFitting an LDA100 xpTidying LDA output100 xpComparing LDA output100 xpInterpreting topics50 xpNaming three topics100 xpNaming four topics100 xpWrap-up50 xp
PrerequisitesIntroduction to the Tidyverse
Assistant Professor of Marketing, BYU Marriott School of Business
Marc's research is focused on applications of Bayesian inference in marketing, including choice modeling and text analysis. He teaches courses in survey research, conjoint analysis, marketing analytics, and statistical modeling.