ISL
Introduction to Statistical Learning
OUTLINE A
- Introduction
- Statistical Learning
- Linear Regression
- Classification
- Resampling Methods
- Linear Model Selection and Regularization
- Moving Beyond Linearity
- Tree-Based Methods
- Support Vector Machines
- Unsupervised Learning
OUTLINE B
1. Introduction
2. Statistical Learning
- What is Statistical Learning?
- Assessing Model Accuracy
- LAB: Intro to R
- EXERCISES
3. Linear Regression
- Simple Linear Regression
- Multiple Linear Regression
- Other considerations in the Regression Model
- The Marketing Plan
- Comparison of Linear Regression with K-Nearest Neighbors
- LAB: Linear Regression
- EXERCISES
4. Classification
- An Overview of Classification
- Why Not Linear Regression?
- Logistic Regression
- Linear Discriminant Analysis
- A Comparison of Classification Methods
- LAB: Logistic Regression, LDA, QDA and KNN
- EXERCISES
5. Resampling Methods
- Cross-Validation
- The Bootstrap
- LAB: Cross-Validation and the Bootstrap
- EXERCISES
6. Linear Model Selection and Regularization
- Subset Selection
- Shrinkage Methods
- Dimension Reduction Methods
- Considerations in High Dimensions
- LAB 1: Subset Selection Methods
- LAB 2: Ridge Regression and the Lasso
- LAB 3: PCR and PLS Regression
- EXERCISES
7. Moving Beyond Linearity
- Polynomial Regression
- Step Functions
- Basic Functions
- Regression Splines
- Smoothing Splines
- Local Regression
- Generalized Additive Models
- LAB: Non-Linear Modeling
- EXERCISES
8. Tree-Based Methods
- The Basics of Decision Trees
- Bagging, Random Forests, Boosting
- LAB: Decision Trees
9. Support Vector Machiens
- Maximal Margin Classifier
- Support Vector Classifiers
- Support Vector Machines
- SVMs with More than Two Classes
- Relationship to Logistic Regression
- LAB: Support Vector Machines
- EXERCISES
10. Unsupervised Learning
- The CHallenge of Unsupervised Learning
- Principal Components Analysis
- Clustering Methods
- LAB 1: Principal Components Analysis
- LAB 2: Clustering
- LAB 3: NC160 Data Example
- EXERCISES
CHAPTER OUTLINES
Chapter 1: Introduction
Chapter 2: Statistical Learning
- What is Statistical Learning?
- Why Estimate F?
- How do we estimate F?
- The Trade-Off between Prediction Accuracy and Model Interpretability
- Supervised versus Unsupervised Learning
- Regression Versus Classification Problems
- Assessing Model Accuracy
- Measuring the Quality of Fit
- The Bias-Variance Trade-Off
- The Classification Setting
- LAB: Intro to R
- EXERCISES
2.1: What is Statistical Learning?
TL;DR: A set of approaches for estimating f (Where f is a function of X, our predictors/input variables, that equal Y our output variable)
- Why Estimate F?
- Prediction
- f can be a black box
- EX: We want to see if this patient’s blood sample will tell us if this person is at high risk for something
- Who will respond positively to a mailing?
- Inference
- When we want to understand the relationship between X and Y
- How Y changes as a function of X
- f cannot be a black box
- Which predictors are associated with the response?
- What is the relationship between the response and each predictor?
- Can the relationship between Y and each predictor be adequately summarized using a linear equation or is the relationship more complicated?
- What effect will changing the price of a product have on sales?
- How much extra will a house be worth if it has a view of the river? TL;DR – Linear is good for inference, non-linear is better for prediction (and worse for interpretability)
- Prediction
- How do we estimate F?
- Parametric Methods
- Trying to fit to a linear model
- DEF: reduce the problem of estimating f down to one of estimating a set of PARAMETERS
- Non-parametric Methods
- DEF: Seek an estimate of f that gets as close to the dta points as possible without being too rough or wiggly
- Large number of observations needed
- Parametric Methods
- The Trade-Off between Prediction Accuracy and Model Interpretability
- Inflexible == more linear == less accurate, easier to interpret
- Flexible == non linear (think svms) == more accurate, harder to interpret
- Supervised versus Unsupervised Learning
- SUPERVISED: For each observation of the predictor measurements (xi), there is an associated response measurement (yi)
- PREDICTION: Accurately predict the response for future observations
- INFERENCE: Better understand the relationship between the response and predictors
- UNSUPERVISED: “Flying blind” – it’s not possible to fit a linear regression bc we don’t have a response variable that can “supervise” our analysis.
- SUPERVISED: For each observation of the predictor measurements (xi), there is an associated response measurement (yi)
- Regression Versus Classification Problems
2.2: Assessing Model Accuracy
- Measuring the Quality of Fit
- The Bias-Variance Trade-Off
- The Classification Setting
- The Bayes Classifier
- K-Nearest Neighbors