Professor Colin Cameron Bainer 1132 email@example.com
Tuesday Thursday 8.00 - 9.50 am SSH 1113 (Blue Conference Room)
Pre-requisites: The listed pre-requisite is Econ / ARE 240D. The essential pre-requisite is Econ / ARE 240D.
Course Goals: The Spring 2016 course includes a survey of
Statistical Learning Methods. This will cover many of the methods
very briefly. The most important for econometrics include
cross-validation, Lasso and regression trees. Following the
survey the course will consider their use in causal econometrics
research. The remainder of the course covers various topics.
Brief Course Outline:
Classes 1-10 Statistical learning
Classes 11-12 Statistical learning for causal econometrics
Classes 13-16 Bayesian analysis and multiple imputation
Classes 17-20 Further topics (most likely including clustering)
For statistical learning the main text is an undergraduate level
ISL: Gareth James, Daniela Witten, Trevor Hastie and Robert Tibsharani (2013), An Introduction to Statistical Learning: with Applications in R, Springer.
A free legal pdf is at http://www-bcf.usc.edu/~gareth/ISL/ and a $25 hardcopy can be obtained via http://www.springer.com/gp/products/books/mycopy
Supplementary material on statistical learning will come from the
graduate level book
ESL: Trevor Hastie, Robert Tibsharani and Jerome Friedman (2009), The Elements of Statistical Learning: Data Mining, Inference and Prediction, Springer.
A free legal pdf is at http://statweb.stanford.edu/~tibs/ElemStatLearn/index.html and a $25 hardcopy can e obtained via http://www.springer.com/gp/products/books/mycopy
Detailed Course Outline:
Classes 1-2 Introduction to Statistical Learning
Statistical learning overview (ISL chapters 1, 2.1-2-2)
Getting started in R (ISL chapter 2.3 and http://cameron.econ.ucdavis.edu/R/R.html)
Linear Regression (ISL chapter 3)
Cross-validation (ISL chapter 5)
Classes 3-5 Linear Model Selection and
Subset selection (ISL chapter 6.1)
Ridge Regression, Lasso, and LARS (ISL chapter 6.2 and ESL pp.73-79, 86-93)
Principal Components and Partial Least Squares (ISL chapter 6.3)
High-dimensional Data (ISL chapter 6.4)
Classes 6-7 Flexible Regression Models
Polynomials, Step Functions and Basis Functions (ISL chapter 7.1-7.3)
Splines (ISL chapter 7.4-7.5)
Local Regression (ISL chapter 7.6)
Generalized Additive Models (ISL chapter 7.7)
Regression Trees (ISL chapter 8.1)
Bagging, Random Forests, Boosting (ISL chapter 8.2)
Classes 8-9 Classification and Unsupervised Learning
Logistic Regression and Discriminant Analysis (ISL chapter 4.1-4.5)
Support Vector Machines (ISL chapter 9.1-9.3)
Unsupervised Learning (ISL chapter 10.1-10.3)
Class 10 Midterm exam
Classes 11-12 Statistical Learning in Econometrics
Double Selection Lasso Belloni, Chernozhukov and Hansen (JEP Spring 2014 pp.29-50)
Big Data for Econometrics Varian (JEP Spring 2014 pp.3-28)
Recursive tree partitioning Athey and Imbens (2015)
Classes 13-17 Bayesian Methods and Multiple
Bayesian Methods Cameron and Trivedi (2005), Microeconometrics: Methods and Applications, Chapter 13.1-13.6 plus notes provided
Multiple Imputation Cameron and Trivedi (2005), Microeconometrics: Methods and Applications, Chapter 13.7, 21.7-27.9
Assignments, data, etc will be posted at the course website at Smartsite under Resources.
Assignments 50% Due Thursdays April 7, 21; May 3, 17; June 2.
Midterm 25% Thursday April 28
Final 25% Wednesday June 8 10.30am-12.30pm Material after midterm.
Assignments must be handed in on time, so solutions can be
discussed in class and distributed in a timely manner.
No credit for late assignments. All must be done.
Academic integrity is required. What is academic integrity? See the UCD Student Judicial Affairs website http://sja.ucdavis.edu/
As an exception to their rules, I permit some collaboration with other students in doing assignments, but the work handed in must be your own. Each person must create their own Stata output and write up their own answers. And you are to write on your assignment the name of the person(s) you worked with.
Exams will be closed book. The final exam is comprehensive.