Machine Learning with Big Data for Social Scientists

The course has two core topics:

• constructing and managing large datasets in R
• machine learning, with specific focus on providing analytics from large datasets. 

For the first topic of big data, we start by asking: What is big data? Why it is difficult to work with? We then learn the best solutions in R for working with big data depending on the size of data, from locally stored data objects to databases hosted on the cloud. 

For the second topic of machine learning, we will learn basic concepts such as:

1. problem definitions
2. objective function
3. bias-variance tradeoffs
4. parameter tuning. 

Social scientists have traditionally emphasised the explanation as the primary purpose of statistical analysis. Machine learning has an overlapping but evidently different orientation. By contrasting the inference-based approach and prediction-focused approach, you get to understand the fundamental ideas of machine learning. The application of machine learning techniques to various analytical tasks in social sciences will follow these theoretical discussions.

The daily topics are listed below.

Day 1

Data Management in R
Discussions on big data, what it is and why it's difficult to work with. Set up of computation environments in the cloud is covered, cloud computing concepts are introduced and why the shift to cloud computing is occuring. To conclude day 1, you will work with large data in R, starting from R objects (dataframe, tibble, etc), databases, and distributed-computing framework. 

Day 2

Databases & Machine Learning Basics
Learn the methods for handling big data using various types of databases, which unload the data from R working space while keeping it highly accessible. Explore the basics of relational databases, such as concepts, rules, and design followed by the syntax of SQL query. Participate in general discussions on the logic of machine learning to understand the basics.

Day 3

Regression
Revisting linear regression, learn how to intrepret the regression problem in the machine learning framework, covering cases where the outcome is a continious quantity. Explore the issue of variable selection which is frequently faced with big data with numerous input features. Learn about typical shrinkage methods such as Ridge regression and LASSO. The concept of the resampling method will also be introduced.

Day 4

Classification Methods 1
Moving to cases where the outcome is categorical, you will participate in discussions on the supervised classification where the machine learning algorithms are applied to predict known values of outputs. Using a classification method familar to social scientists, learn how to evaluate the models in machine learning framework. A more detailed discussion of model performance evaluation will be provided.

Day 5

Classification Methods 2
Continuing on from Day 4, the focus is on various tree-based methods. Starting with a simple tree, moving on to more sophisicated methods such as random forest and boosting. To finish, we will revist the issue of data size and get an overview of the methodology for distributed computing, which has the capability of getting insights from big data as a whole.

How the course will work online

The course provides approximately 5 hours of pre-recorded lectures as well as an online forum on Slack where the Instructor and students can freely discuss the lecture materials and coding. 

Approximately two hours of each day will be an online seminar, where we will learn how to apply the concepts and knowledge gained from pre-course lecture materials through Q&A and the live lab work. 

In the live lab, you will be given several coding tasks, and asked to code along with the Instructor. Some tasks are left as homework, which will be discussed on the online forum and during the following day’s live lab. 

You’ll also be able to advance-book one-to-one consultations with the Instructor during office hours scheduled in advance. 

You will learn how to work with cloud computational workspace using R as a primary statistical software with RStudio server and Google Colab. Example scripts and assignments are distributed through github so you can learn how to use online version control systems for collaboration and research accountability.

The course assumes you have some familiarity with R statistical language and can conduct basic data handling in R (opening data files, working with data frames). If you don’t have this, take the week-one course Introduction to R. 

You should also have basic knowledge of standard statistical analysis in social science, such as linear regression and hypothesis testing. These are covered in courses Introduction to Inferential Statistics and Big Data Collection and Management in R.

The course consists of two activities which are to be completed before Day 1.

• Reading - consisting mostly of textbook type materials. Approximately two hours for each day of the course. (10 hours)
• Lectures - pre-recorded videos accompanied by supporting materials. A one hour video for each day of the course. (5 hours)