TBA
summary
Demonstrates why a line is the practical backbone of prediction: simulates a 40-point cloud by adding Gaussian noise to a true line, fits a least-squares line in R and shows predictions that sit consistently off the true model because of the bias–variance trade-off. Gives concrete wins (Netflix used regressions to greenlight House of Cards) and concrete failures (Night Capital lost about $400M from opaque algorithms), and recommends Peter Flach’s Machine Learning and using R/RStudio for analysis.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 3:29Speaker background
- 5:55Statistical lines
- 7:02What vs why happened
- 9:14Regression purpose
- 12:22Know your data
- 14:52Simple visualizations
- 16:37ML output struggle
- 18:44Intuition vs complexity
- 19:50Linear regression explained
- 26:11Non-straight line reality
- 27:54Modeling with linear equations
- 31:02Minimizing distance
- 33:53Predictions from linear models
- 36:20Bias variance trade-off
- 39:17Linear models for non-linear
- 41:43Extending linear regression
- 43:08Dangers of oversimplified models
- 44:53Learning statistics best practices
- 46:16Data science demystified
