The Analytics Power Hour #267: Regression? It Can be Extraordinary! (OLS FTW. IYKYK.) with Chelsea Parlett-Pelleriti
24 snips
Mar 18, 2025 Chelsea Parlett-Pelleriti, a statistician and data scientist from Recast, returns to dive deep into the fascinating world of linear regression. She simplifies complex concepts like feature engineering, multicollinearity, and overfitting, making statistics accessible. The discussion explores how statistical models can improve marketing efforts and the distinction between predictive and inferential analysis. They also share lighthearted anecdotes about pet safety and social media experiences, blending humor with insightful analytics.
AI Snips
Chapters
Transcript
Episode notes
Regression's Importance
- Linear regression is a fundamental tool, even for complex models like MMM.
- It provides a foundation for understanding relationships between variables.
Start Simple
- Start with the simplest method to solve problems.
- Use linear regression if the problem fits its parameters and is easily communicated.
Supervised vs. Unsupervised
- Linear regression works best with supervised questions where the answer is known for some of the data.
- Unsupervised problems like customer segmentation don't fit linear regression.