Geometric Interpretation of Linear Regression

In this lesson

1. 0:00What we're buildingIntroduce the regression problem and what a linear model is choosing between.
2. 2:41What the data looks likeInspect the data on a graph before deciding how to model its relationship.
3. 3:33The simplest possible modelBuild intuition for using a straight line as a simple predictive model.
4. 6:17The model and its two parametersIdentify the model’s two parameters and understand how they change the fitted line.
5. 9:02Building scoring metricConstruct a score from prediction errors and explain why the errors are squared rather than simply added.
6. 15:58Training it: gradient descentFollow gradient descent step by step to see what training the model actually means.
7. 23:26Reading the modelInterpret the fitted coefficients and distinguish what they tell you from what they do not.
8. 26:38How good is it? - R2 explainedUse R squared to assess the fit and consider when that score can give a misleading impression.
9. 31:22Where it breaksExamine limitations of a straight-line model and recognise situations where its predictions can mislead.
10. 34:10In codeConnect the geometric explanation of linear regression to its implementation in code.
11. 38:33RecapReview the path from data and model parameters through training, interpretation, and evaluation.