Predictive Analytics: The Power To Predict Who Will Click, Buy, Lie, Or Die (2013) - Plot & Excerpts
Correlation far more important than causation, so long as the correlations are good. Machine learning proceeds from induction, reasoning from detailed facts to general principles. Though not mentioned by name, the Canadian Tire spending analysis that included Sharky's was referenced!Chapter on Ensemble models and IBM's Watson playing Jeopardy! was very interesting. The language and idiom barrier that had to be overcome to answer the questions was massive, researchers used many models in concert. Ensemble models combine evidence and are specialized for each type of question. Then other levels of predictive models are applied to increase confidence, and Watson was taught to answer when confidence was lower when needed. The biggest competitive question was whether Watson had a speed advantage when buzzing in. Final chapter about Persuasion Effect and Uplift modeling also very interesting. Legacy solutions of churn modeling and response modeling cannot show causality. Uplift modeling involves a contacted set and a control set, and compares response rates among similar cohorts. This kind of modeling was used in Obama's reelection campaign. A good motivator for harnessing technology for financial gains; the author's perspective is mostly presented as a marketing consultant. Because of this much of the content hinges on case studies of client's who direct PA for maximizing ROI of Spam campaigns. The best chapter, in my opinion, is spent detailing how Watson, IBM's Jeopardy-winning computer leverages ensemble modeling (and a big database of information) for the sake of AI.
What do You think about Predictive Analytics: The Power To Predict Who Will Click, Buy, Lie, Or Die (2013)?
Easy read and excellent approach of Siegel to explain predictive analytics on an introductory level.
—jesse159
Not bad. Lots of interesting different examples of where predictive analytic was used.
—anime4182
Heavy on flowery language, light on details; an exciting overview of the power of PA.
—misinmyname
If you want to know anything (and most everything) about PA, read this!
—jake_maxwell
