Physicist Niels Bohr observed: “Prediction is very difficult,
especially if it’s about the future!” Despite that warning,
scientists, politicians, journalists, doctors, economists--professionals in
nearly every knowledge domain--try to anticipate what’s in
store for us using everything from divination, “gut feelings,” and
“educated guessing” to complex statistical modeling, machine learning,
and artificial intelligence (AI).
Data science expert and educator Stylianos Kampakis has written a
valuable and instructive book on prediction, subtitled the history
and future of data science and artificial intelligence, not to
tell readers what might happen next but to explain the tools of
prediction under uncertainty: their history, utility, and potential
impact. He calls uncertainty an unavoidable “enemy” that humanity
has been fighting to control using logic, science, and reason,
recognizing that it can’t be eliminated, only understood and reduced.
The opening chapter presents a short history of uncertainty, citing
the development of statistical science; noting that there are
different kinds and levels of uncertainty; and providing examples from
weather, polling, medicine, economics, life choices, and investing.
Subsequent chapters discuss the predictive power and challenges of
logic, inductive reasoning, probability, and hypothesis testing,
including an obligatory yet readable review of Bayesian versus frequentist
analyses.
The second half of the book tends more philosophical, considering
the nature of causality, human perception and interpretation of
uncertainty, and the limits of prediction, including the shortcomings
of machine learning, simulations, and trust technologies like
blockchain. Throughout, the author highlights views about probability and uncertainty from historical
practitioners and thinkers, including Locke, Bacon, Shannon, Fisher, Laplace, Gauss, Pascal, and Tukey,
to name but a few luminaries in this domain.
Understanding uncertainty, and how to describe it and manage it using
modern data analytics tools and methods, is increasingly critical
in today’s social, economic, and scientific endeavors. Kampakis’s
book clearly and readably covers the essence of uncertainty and
the human efforts to address it, written for both professional data
scientists and anyone attempting to predict life’s unknowable
and unexpected outcomes.
More reviews about this item: Amazon