Overview
Discussion Board Activity: kNN Regression vs. Multiple Linear Regression
Background
In our classification module, we used accuracy as a simple and intuitive measure of model performance. However, accuracy alone can be misleading, especially when classes are imbalanced. To become thoughtful analysts, it’s important to understand why accuracy is often insufficient and what questions we need to ask beyond the accuracy value.
Preparation
Before posting, take a moment to reflect on:
Situations from class where the confusion matrix revealed insights that accuracy alone could not
Real-world problems (business, healthcare, operations, finance) where misclassifying one class is more costly than the other
Cases where a model could achieve a high accuracy by simply predicting the majority class
Why analysts often look beyond accuracy to understand a model’s strengths and weaknesses
You may want to briefly review the concept of class imbalance, error types (false positives vs. false negatives), and what information each cell in the confusion matrix conveys.
Suggested Discussion Points
Why accuracy can be misleading, especially with imbalanced datasets
An example where a model achieves high accuracy but performs poorly on a critical class
How different types of misclassification (false positives/false negatives) have different real-world costs
Why the confusion matrix provides a deeper understanding of performance than accuracy alone
A situation from your experience (or a hypothetical one) where relying only on accuracy could lead to a bad decision