From the course: Machine Learning and AI Foundations: Classification Modeling

Unlock the full course today

Join today to access over 24,000 courses taught by industry experts.

So-called “black box” techniques

So-called “black box” techniques

From the course: Machine Learning and AI Foundations: Classification Modeling

Start my 1-month free trial Buy for my team

So-called “black box” techniques

“

- [Instructor] The modeling approaches that we'll be discussing very greatly and how transparent they are during validation and interpretation. This raises the issue of the so-called black box techniques. A famous example is artificial neural networks. The very same complexities that often allow for very accurate models can prevent you from sharing a story about the model. Sometimes something as basic as a key driver analysis. Just simply a list of predictors in order of importance can seem elusive. Another easy to imagine example is an ensemble of trees. A single decision tree is among the most transparent, but a so-called random forest is really a collection of trees. Essentially an average of 100 or even 1,000 models which then becomes difficult or impossible to interpret. The truth is that virtually all of these techniques are in a continuum from transparent to opaque. Perhaps the most transparent that we will see in this course is logistic regression, where all of the inputs can…

Contents