Machine learning is programming computers to build predictive models for making inference from samples for accurate out-of-sample predictions and decisions. As an applied economist, or researcher in a related field, if you feel that you need to learn these contemporary machine learning methods, implement them into your research, or looking for articles applied these methods to incorporate into the courses that you are teaching, this course is for you. In this course, participants will learn the fundamentals of machine learning and their implementation, recalling the fundamental statistical concepts at the heart of modern learning techniques. Participants will learn the differences between causal and predictive analyses and their relative merits, as well as their use in applied social sciences.
The course will combine both real data and theoretical background to enable researchers to gain practical experience in analysing a wide variety of data and econometric problems. It will also discuss how contemporary approaches in applied econometrics can be used to answer important questions in Economics. Participants will be provided a wealth of research papers and sources which apply the techniques being taught. Applications covered during the course will include the fields of labour, development, industrial organisation, health, macro economics and finance.
By the end of the course, participants will be able to understand various concepts intensively used in the machine learning literature such as cross-validation, bootstrapping, boosting, bagging, optimization routines, and identify common techniques, such as CART, Random Forest, MARS, GAM, Lasso, most suitable for their research questions and data and their natural extensions to causal inference with observational data.
The participants will be expected to be familiar with concepts and practices of traditional multivariate regression analysis, and related estimation and causal methods.
All examples will be given in R or Stata, so you should also have a working knowledge of these software (which can be replaced by an advanced knowledge of another programming language, such as Python, to adapt the codes and exercises on your own). Participants are invited to bring their own laptop with R-Studio and Stata installed on it.
Lecturers
The Schedule
The course will follow the book, Statistical Learning by Machines in R (forthcoming - 2020). The online version of the related notes will be provided to the participants. Data sets and R code will be available through a supporting website. The first sessions will start at 10am. Following the breaks between 12:30pm and 2:00pm, the second sessions will start at 2pm and end at 4:30pm. For those who would like additional training in R, there will be R Labs the first three day between 5:30pm and 7pm. All times are Atlantic Time zone.
10am ADT (UTC -3) / 6:00am PDT / 9:00am EDT / 2:00pm London / 4:00pm Istanbul
July 20, Monday
Session 1 (10am - 12:30pm): Causal vs. Predictive Models, Translation of Concepts Used in Machine Learning for Social Scientists
Session 2 (2pm - 4:30pm): Overfitting, Model Selection, and Variance-Bias Trade-Off.
R-Session (5:30pm - 7pm): Additional training in R - voluntary
July 21, Tuesday
Session 1 (10am- 12:30pm): Regressions and Linear Classifiers, Nonparameteric Models, kNN.
Session 2 (2pm -4:30pm): Grid Search with Cross-Validation and Bootstrapping.
R-Session (5:30pm - 7pm): Additional training in R - voluntary
July 22, Wednesday
Session 1 (10am- 12:30pm): Classification and Regression Trees, Random Forests, Boosting, Bagging.
Session 2 (2pm -4:30pm): Penalized Regression Models: Lasso, Ridge, Elastics Net, and Adaptive Lasso.
R-Session (5:30pm - 7pm): Training in Data Revisions and Now-casting of macroeconomics series by Andrea Guisto,PhD
July 23: Presentations of Speakers
10am - 10:55am : Juri Marcucci: Machine Learning in Macroeconomics (presentation) (video)
11am - 11:55am : Arthur Charpentier: Machine Learning in Actuarial Science & Insurance (presentation) (video)
12pm - 12:55pm : Arthur Spirling : Machine Learning in Embeddings Representations (presentation) (video)
1pm - 1:55pm : Kathy Baylis: Machine Learning in Agricultural Economics (presentation) (video)
2pm - 2:55pm : Stefan Wager : Machine Learning in Causal Inference (presentation) (video)
July 24: Presentations of Speakers
10am - 10:55am : Stan Matwin : Machine Learning and Economics: a two-way street
11am - 11:55am : Mehmet Caner : Machine Learning in Econometrics (presentation) (video)
12pm - 12:55pm : Anders Bredahl Kock : Machine Learning in Model Selection (presentation) (video)
1pm - 1:55pm : Dario Sansone: Machine Learning in Education and Development Economics (presentation) (video)
2pm - 2:55pm: Patrick Baylis :Temperature and Temperament: Evidence from Twitter (presentation) (video)
Speakers:
(Alphabetical order)
Professor, Université du Québec à Montréal
Machine Learning in Actuarial Science & Insurance
Professor of Politics and Director of The Center for Data Science , NYU
Machine Learning in Political Science
Postdoctoral Scholar Vanderbilt University
Machine Learning in Education and Development Economics
Economist at the Research Department , Bank of Italy
Machine Learning in Macroeconomics
Professor, Agricultural and Consumer Economics, University of Illinois
Machine Learning in Agricultural Economics
Thurman-Raytheon Distinguished Professor of Economics, NC State
Machine Learning in Econometrics
Assistant Professor Vancouver School of Economics
Temperature and Temperament: Evidence from Twitter
Professor and Director, Institute for Big Data Analytics, Dalhousie University
Machine Learning and Data Privacy
Assistant Professor of O.I.T. and Statistics at the Stanford Graduate School of Business
Machine Learning in Causal Inference
Please, click webinar for detailed information about speakers
FEES
1. Advanced Ph.D.s, Postdocs and temporary university researchers: $200 CAD.
2. Faculty and researchers: $300 CAD. (Participants from Developing Countries :$200 CAD)
3. Others: $600 CAD.
Fee includes the course related documents. Lectures and speakers are receiving no monetary benefit from both event. All fees will be used to support MLportal's Research Assistants and projects. If you are under financial stress due to COVID19, Please send us an email.
RSVP deadline: July 15th, 2020.
Fees Payment deadline: July 19th, 2020.
The Online Lecture is conditional to the recruitment of a minimum of 10 participants. In case the number of applicants is beyond the manageable level, we will limit the number of participants.
ENROLLMENT
The course and the webinar series will be live at Zoom. The details will be provided via email.
To register Please fill RSVP form.
This course is mainly designed for researchers who are using observational data to investigate the effect of a variable on an outcome. The participants will be expected to be familiar with concepts and practices of traditional multivariate regression analysis, and related estimation and causal methods. Therefore, if you are not associated with a university or a research institution, please attached a short CV to your email.
Please fill RSVP form if you only want to attend webinars, which is free.
For any questions mlportalns@gmail.com