Abstract
Advances in digital technology have led to the digitization of everyday activities of billions of people around the world, generating vast amounts of data on human behavior. A parallel trend has been the emergence of computational methods and analysis techniques needed to deal with these new sources of behavioral data—which tend to be more unstructured, of much larger scale, and noisier. As they are recent and emerging developments, many behavioral scientists and practitioners may be unaware or unfamiliar with these recent developments and opportunities to extract behavioral insights. The main objective of this article is to discuss machine learning methods for researchers and practitioners interested in addressing customer-relevant questions using new secondary data sources that are publicly available, such as data from crowdfunding, video streaming, crowdsourcing, and social media platforms. This article offers a primer on the application of computational social science for understanding consumer data for researchers and practitioners.
TopBackground
As both the availability of large-scale behavioral data and computational analysis methods are recent and emerging developments, many behavioral scientists and practitioners may be unaware or unfamiliar with (1) new sources of secondary data and types of data that are available to extracts insights about consumer behavior, and (2) new analysis techniques to study consumer behavior at scale. Therefore, motivated by these recent developments and opportunities, the main objective of this book chapter is to discuss computational methods (specifically, machine learning methods) for researchers and practitioners interested in addressing customer-relevant questions using new secondary data sources that are publicly available. This chapter offers a primer on the application of computational social science for understanding consumer data for both researchers and practitioners.
Key Terms in this Chapter
Unstructured Data: Data that is not predefined through structured, number-based systems, such as texts, sounds, visual images, or videos.
Natural Language Processing: A field of study to derive computational approaches to process and analyze large amounts of textual data on human (natural) language.
Machine Learning: An approach to derive computer algorithms and statistical models that can learn to improve their performance based on use of data, without explicit instructions (data-generative models).
Text Mining: The process of identifying, retrieving, and preparing unstructured textual data into a structured format for data analysis.
Automatic Speech Recognition: A field of computer science and class of methods which enables the recognition and translation of spoken language into written text that is processed by computer systems.
Structured Data: Data that is predefined through number-based formats and typically considered quantitative data, such as total revenues, counts of likes on a social media post, number of products sold, customer satisfaction ratings, and so on. Their measurement scales could be nominal, ordinal, interval, and ratio.
Computer Vision: A field of computer science and class of methods which allow computer systems to process and make sense of visual images, akin to human vision system.
Customer-Relevant Data: Data in relation to information that customers are exposed to in the marketplace as well as data generated by customer’s behavior.
Computational Linguistics: A field of study and class of methods which allow computer systems to process and make sense of natural (human) language.
Waveform Analysis: Analysis of the acoustic features of the raw waveform file containing audio (sound and speech).
Computational Social Science: An emerging field of study which leverages computational techniques to study social and behavioral phenomena of individuals, groups, and societies.