9  Alternative Algorithms (TODO)

As the consummate showman, P.T. Barnum is often quoted as saying “Leave them wanting more”. Unfortunately, statistics professors have less of a flare for drama. Introductory statistics courses will typically introduce a few types of models (for example, linear and perhaps logistic regression), and that’s a wrap. It’s often until students start taking the subsequent courses that they are exposed to the true limitations of previous techniques and taught to demand more.

This chapter attempts to flip that paradigm by briefly surveying a broad number of modeling techniques. The goal is not to go into all of the rigorous deals that one should understand to use these models. Instead, we hope to build a “mental toolbox” of techniques so that you know where to focus your study when you encounter a problem in the real world.

9.1 Not Modeling

9.1.1 First Principles

9.1.2 Simple Analyses

9.2 Extending Linear Regression

9.2.1 Modeling Binary Outcomes

9.2.2 Modeling Counts

9.2.3 Modeling Time Until an Event

9.2.4 Modeling Repeated Measures on a Population

9.2.5 Modeling Observations in a Nested Hierarchy

9.3 Causal Analysis Patterns

Similar to https://emilyriederer.netlify.app/post/causal-design-patterns/

9.4 Special Data Types

9.4.1 Duration Analysis

9.4.2 Time & Space Data

9.5 Bayesian Methods

9.6 Simulation Methods

9.6.1 Agent-Based

9.6.2 Discrete Event

9.7 Clustering (beyond K-Means)

9.7.1 Density-Based

9.7.2 Mixture Models