Dark data : why what you don't know matters

상세정보

자료유형	단행본
서명/저자사항	Dark data: why what you don't know matters / David J. Hand.
개인저자	Hand, D. J. (David J.), 1950- author.
발행사항	Princeton, New Jersey: Princeton University Press, [2020]
형태사항	xii, 330 p. : Illustrations ; 23 cm
ISBN	9780691234465 : 9780691182377 069118237X
서지주기	Includes bibliographical references and index.
내용주기	Part 1. Dark data: their origins and consequences -- Chapter 1. Dark data: what we don't see shapes our world. Ghost of data -- So you think you have all the data? -- Nothing happened, so we ignored it -- Power of dark data -- All around us -- Chapter 2. Discovering dark data: what we collect and what we don't -- Dark data on all sides -- Data exhaust, selection, and self-selection -- From the few to the many -- Experimental data -- Beware human frailties -- Chapter 3. Definitions and dark data: what do you want to know? -- Different definitions and measuring the wrong thing -- You can't measure everything -- Screening -- Selection on the basis of past performance -- Chapter 4. Unintentional dark data: saying one thing, doing another -- Big picture -- Summarizing -- Human error -- Instrument limitations -- Linking data sets -- Chapter 5. Strategic dark data: gaming, feedback, and information asymmetry -- Gaming -- Feedback -- Information asymmetry -- Adverse selection and algorithms -- Chapter 6. Intentional dark data: fraud and deception -- Fraud -- Identity theft and internet fraud -- Personal financial fraud -- Financial market fraud and insider trading -- Insurance fraud -- And more -- Chapter 7. Science and dark data: the nature of discovery -- Nature of science -- If only I'd known that -- Tripping over dark data -- Dark Data and the big picture -- Hiding the facts -- Retraction -- Provenance and trustworthiness: who told you that? Part II. Illuminating and using dark data -- Chapter 8. Dealing with dark data: shining a light -- Hope! -- Linking observed and missing data -- Identifying the missing data mechanism -- Working with the data we have -- Going beyond the data: what if you die first? -- Going beyond the data: imputation -- Iteration -- Wrong number! Chapter 9. Benefitting from dark data: refraining the question -- Hiding data -- Hiding data from ourselves: randomized controlled trials -- What might have been -- Replicated data -- Imaginary data: the Bayesian Prior -- Privacy and confidentiality preservation -- Collecting data in the dark -- Chapter 10. Classifying dark data: a route through the maze -- Taxonomy of dark data -- Illumination. Dark data : their origins and consequences -- Illuminating and using dark data.
요약	"Data describe and represent the world. However, no matter how big they may be, data sets don't - indeed cannot - capture everything. Data are measurements - and, as such, they represent only what has been measured. They don't necessarily capture all the information that is relevant to the questions we may want to ask. If we do not take into account what may be missing/unknown in the data we have, we may find ourselves unwittingly asking questions that our data cannot actually address, come to mistaken conclusions, and make disastrous decisions. In this book, David Hand looks at the ubiquitous phenomenon of "missing data." He calls this "dark data" (making a comparison to "dark matter" - i.e., matter in the universe that we know is there, but which is invisible to direct measurement). He reveals how we can detect when data is missing, the types of settings in which missing data are likely to be found, and what to do about it. It can arise for many reasons, which themselves may not be obvious - for example, asymmetric information in wars; time delays in financial trading; dropouts in clinical trials; deliberate selection to enhance apparent performance in hospitals, policing, and schools; etc. What becomes clear is that measuring and collecting more and more data (big data) will not necessarily lead us to better understanding or to better decisions. We need to be vigilant to what is missing or unknown in our data, so that we can try to control for it. How do we do that? We can be alert to the causes of dark data, design better data-collection strategies that sidestep some of these causes - and, we can ask better questions of our data, which will lead us to deeper insights and better decisions"--
해제	Provided by publisher.
일반주제명	Missing observations (Statistics) Big data. Observations manquantes (Statistique) Données volumineuses. Big data. Missing observations (Statistics)
분류기호(DDC)	519.5
언어	영어