---
title: Predictive models demystified
---

Discover 5 practical ways to evaluate risk, validate predictions, and pick the right model for your goals. ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­    ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­  

| View in browser September 2026 The Analytics Advantage In addition to my day job, I’ve coached many of my kids’ sports teams, including my son’s middle school baseball team. During one playoff game, we were tied in the final inning with the bases loaded and one out. Our strongest hitter was due up, but to everyone’s surprise, I sent in a player who rarely got a hit. To many, the decision probably seemed crazy. But in that situation, we didn’t need a home run. We needed someone who would avoid swinging at bad pitches and make contact if he swung. Finding the “best” option wasn’t about a single statistic; it was about understanding tradeoffs and choosing the player who best fit the moment. The same is true in analytics. In this month's issue, we explore why model accuracy isn't always the right measure of success, how model validation helps ensure predictions stand up to real-world use, and how structural equation models can help uncover causal relationships and answer the “why” behind complex situations. We also share practical tips for selecting the predictive modeling approach that's right for your goals. And if you’re curious how that game ended, we won on a four-pitch walk! Enjoy this month’s issue.   Larry LaRusso Editor, The Analytics Advantage Why your most accurate model might still be costing you It may sound counterintuitive, but the most accurate model isn’t always the best one. In this blog, Olivia Lippincott explains why some prediction errors are far more costly than others and how that affects the way classification models should be evaluated. Learn why false positives and false negatives should be treated differently, how to weigh risk and reward, and how to optimize your models for better business outcomes, not just better accuracy. (5-minute read) Read the blog Expert tips for picking the right predictive model With so many predictive modeling techniques available, choosing the right one can feel overwhelming. The “best” model ultimately depends on what you’re trying to achieve. In this 30-minute webinar, we take the guesswork out of selecting the model that’s right for you. Using a practical case study, learn how different models are fitted, compared, and refined, giving you a clear roadmap for future projects. (30-minute talk)  View the on-demand webinar Master in a minute Model validation How do you know if your predictive model will perform well on new data? In this lesson from the Statistics Knowledge Portal, learn how model validation (called honest assessment by statisticians) helps determine whether a model will generalize to future observations. We also explore the bias-variance tradeoff, holdout validation, k-fold cross-validation, ROC curves, and common performance metrics used to evaluate predictive models  (2-minute read)    View the lesson Enjoying this newsletter? Forward to a friend or subscribe here. A better way to answer 'why': Structural equation models and the science of cause and effect In this era of AI and powerful large language models, answers are often just a click away. But understanding the “why” behind those answers sometimes requires a different approach. In this article, Anne Milley explores how structural equation models (SEMs) help researchers evaluate assumptions, derive meaning, and answer complex questions. Milley also introduces readers to world-renowned expert Dr. Ken Bollen’s recent book, Elements of Structural Equation Models, in which he explores the origins of SEMs, common misconceptions about them, and their strengths and vulnerabilities.  (3-minute read)  Read the article Applying Structural Equation Models to Real-World Problems Data-driven decisions rely not only on trustworthy data but also on analytic tools that support deeper understanding. Structural equation modeling (SEM) provides a flexible, multivariate framework for analyzing complex relationships and answering difficult questions. This on-demand webinar provides an overview of SEM, followed by practical, real-world examples that demonstrate how this approach can provide unique insights that other methods might miss. (30 minute talk)  View the on-demand webinar Top-rated software loved by data pros! See why users trust JMP for breakthrough discoveries. Read user reviews Get your free JMP trial today JMP and all other JMP Statistical Discovery LLC product or service names are registered trademarks or trademarks of JMP Statistical Discovery LLC in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies. Copyright © JMP Statistical Discovery LLC. All Rights Reserved. JMP Statistical Discovery LLC, 920 SAS Campus Drive, Cary, NC, 27513, USA   Your Subscription You are receiving this email because you have subscribed to Analytics Advantage. To stop these emails, visit our unsubscribe page, and enter the email address associated with your subscription: |
| --- |