APPLY THE RIGHT ANALYTIC TECHNIQUE
Applied Predictive Analytics: Principles and Techniques for theProfessional Data Analyst shows tech-savvy business managers anddata analysts how to use predictive analytics to solve practicalbusiness problems. It teaches readers the methods, principles, andtechniques for conducting predictive analytics projects, from startto finish. Internationally recognized data mining and predictiveanalytics expert Dean Abbott provides a practical and authoritativeguide to best practices for successful predictive modeling,including expert tips and tricks to avoid common pitfalls.
This book explains the theory behind the principles ofpredictive analytics in plain English; readers don’t need anextensive background in math and statistics, which makes it idealfor most tech-savvy business and data analysts. Each of thechapters describes one or more specific techniques and how theyrelate to the overall process model for predictive analytics. Thedepth of the description of a technique will match the complexityof the approach, with the intent to describe the techniques inenough depth for a practitioner to understand the effect of themajor parameters needed to effectively use the technique andinterpret the results.
Each of the techniques is illustrated by examples, either uniqueto the task or as part of predictive modeling competitions. Thecompanion website will provide all of the data sets used togenerate these examples, along with links to open source andcommercial software, so that readers can recreate and explore theexamples.
With detailed descriptions of techniques that get results,Applied Predictive Analytics shows you how to:
Choose the proper analytics technique for variousscenarios
Avoid common mistakes and identify the weaknesses of varioustechniques
Mitigate outliers and fill in missing data whennecessary
Interpret predictive models often considered “blackboxes,” including model ensembles
Learn how to assess model performance so the best model isselected
Apply the appropriate sampling techniques for building andupdating models