Applied Text Analytics: A Visual Approach (303129)

Keywords: text mining, data visualization, unstructured data, text analytics

Statistical practitioners across all industries need to find new ways to rapidly exploit the growing unstructured (text) data populating their enterprise. Text analytics has emerged as a powerful method to add to their existing predictive analytics toolkit to uncover structure and insights for business solutions. Colorful word clouds with sizes of words proportional to use frequency are exciting and potentially informative. However, this session goes well beyond the word counts to show how graphics from tabulations, singular value decompositions, cluster analyses, decision trees, and other statistical routines can be used to find common themes, extract information, group like documents and words, and integrate text data with structured predictive analytics. Demonstrations will use examples from government, manufacturing, medical devices, and other industries to illustrate the ease of exploration of common text data sources such as social media, free-text survey results, incident report narratives, and a corpus of documents such as journal articles. We will use a few different software packages including R, SAS Text Miner, and SAS JMP.