Summer Session 2018
Research Design for Data Analytics
From Technique-Centric View Toward Question-Centric View
- Instructor: Jiyong Park, PhD Candidate at KAIST (jiyong.park@kaist.ac.kr)
- This session DOES NOT aim to learn a series of analytics techniques, rather DOES aim to learn what technique applies to what question, pursuing "the right tool for the right question."
- This session covers the research design for data analytics in academic research, rather than practical applications, by integrating applied econometrics, data analytics, and machine learning into a single framework.
- This session is designed with three modules.
- All materials can be downloaded at each page. You can also download the codes on Github.
- This session builds largely upon the following courses or workshops:
- Using Big Data to Solve Economic and Social Problems (Prof. Raj Chetty), Department of Economics, Stanford University
- Workshop on Research Design for Causal Inference, Northwestern University and Duke University
- Summer Institute in Computational Social Science, Princeton University and Duke University
- Wharton PhD Tech Camp 2018 (Bowen Lou, PhD Candidate), Wharton School, University of Pennsylvania
- Machine Learning (Prof. Andrew Ng, Stanford University), Coursera
- Introduction to Machine Learning (장병탁 교수), 컴퓨터공학부, 서울대학교
- There is no prerequisite for this session. Feel free to attend the session selectively.
(Lecture) Research Design for Data Analytics
1st Session
(6/25, 14:00~15:30)
- Two Paradigms of Analytics: Input-Output Framework
2nd Session
(6/28, 14:00~15:30)
- The Right Tool for the Right Question
- Why is Predictive Analytics not Well-Suited for Causal Inference?
- Why is Identification Strategy not Well-Suited for Prediction?
3rd Session
(7/2, 14:00~15:30)
- Causal Inference: Identification Strategy (1)
- Randomized Experiment
- Quasi-Experiment
4th Session
(7/5, 14:00~15:30)
- Causal Inference: Identification Strategy (2)
- Instrument Variable Approach
- Dynamic Panel Model (Internal Instruments)
5th Session
(7/9, 14:00~15:30)
- Reflection on Predictive Analytics
6th Session
(7/12, 14:00~15:30)
- The Art of Prescriptive Analytics: Predictive Analytics + Causal Inference + Optimization
7th Session
(7/16, 14:00~15:30)
- Computational Social Science: Computer Science for Social Science
8th Session
(7/19, 14:00~15:30)
- Economining: Data Mining and Machine Learning for Empirical Research
(Hands-on) Causal Inference with STATA
1st Session
(7/23, 14:00~15:30)
- Replication Project (1) Randomized Experiment
2nd Session
(7/26, 14:00~15:30)
- Replication Project (2) Difference-in-Differences
3rd Session
(7/30, 14:00~15:30)
- Replication Project (3) Instrument Variable Approach
4th Session
(8/2, 14:00~15:30)
- Advanced STATA Programming
(Hands-on) Deep Learning with PyTorch
1st Session
(8/6, 14:00~15:30)
- (Lecture) Deep Learning 101 - Part 1.
2nd Session
(8/9, 14:00~15:30)
- (Lecture) Deep Learning 101 - Part 2.
3rd Session
(8/13, 14:00~15:30)
- Web Data Extraction
4th Session
(8/16, 14:00~15:30)
- Linear Regression in the PyTorch Way
- Neural Network
5th Session
(8/20, 14:00~15:30)
- Convolutional Neural Network (CNN)
- (Application for NLP) Sentence Classification
6th Session
(8/23, 14:00~15:30)
- Recurrent Neural Network (RNN)
- (Application for NLP) Machine Translation (Sequence-to-Sequence)
7th Session
(8/27, 14:00~15:30)
- Deep Reinforcement Learning
- Deep Q-Network (DQN)
8th Session
(8/30, 14:00~15:30)
- Leveraging Deep Learning on Your Computer
- Supervised Learning using Amazon Mechanical Turk
- How to Use Cloud-based APIs
- Running Deep Learning on the Cloud (Amazon Web Service)