Survey researchers' use of machine learning
Frauke Kreuter, Ph.D.
University of Maryland-Mannheim
Trent Buskirk, Ph.D.
University of Massachusetts-Boston
The amount of data generated as a by-product in society is growing fast including data from satellites, sensors, transactions, social media and smartphones, just to name a few. Such unstructured data arise organically also as part of survey data collection processes. Faced the deluge of data, machine learning techniques are often used or called for. This presentation will de-mystify the term machine learning for those not familiar with the way computer scientists approach data analysis task, and give a broad overview over the different techniques often grouped into supervised and un-supervised learning. The goal of this talk is not to dive into the algorithmic details but rather give a high level road-map for researchers as for which techniques to use for what kind of problems, and what process steps to think about.