Applied Data Analysis & Visualization II

Course goals

Learning objectives

After successfully completing this course, you will be able to:

• Understand and explain the different approaches to unsupervised data analysis;
• Given a practical data science problem, select appropriate techniques to tackle this problem;
• Apply various (un)supervised data analysis techniques, including correspondence analysis, principal components analysis, clustering methods, deep learning, structural equation modelling, etc. in R;
• Interpret and evaluate the results of such analyses;
• Explain these evaluations in layman's terms
• Construct appropriate visualisations in connection with each of the data analysis techniques

Content

This course is part of the minor Applied Data Science for Economists.

In the course Applied Data Analysis and Visualization for Economists (ECB2ADAVE) you learned all about supervised learning methods for observable variables. These methods focus on answering questions like what puts former criminals on the right track? How can we prevent heart disease? Can information from social media platforms predict election outcomes? What does a violent brain look like? Hence, these supervised learning methods are statistical methods that are focused on predicting an observable outcome from observable predictors. Also, you learned how to visualise data and analysis results that fit within the concept of supervised learning.

But what if you want to answer these questions for variables that cannot be directly observed, such as subjective well-being, utility, ability, cognitive and non-cognitive skills (human capital)? If research questions concern latent variables, then structural equation modelling applies. And what if you want to answer questions like what type of shoppers do we have on an online shopping site? Or are there different disease types within patients with breast cancer? These datasets are typically characterised by (many) observed variables, but no outcome variable. Here, we want to use techniques from the field of unsupervised learning. Using unsupervised learning and visualisation methods, you will be able to answer questions like can we discover a smaller number of latent dimensions underlying the observed variables or can we discover hidden subgroups among the observations?

During this course, you will actively learn how to apply the main statistical methods in unsupervised, exploratory data analysis and how to use machine learning algorithms and visualising techniques. The course has a strongly practical, hands-on focus: rather than primarily focusing on the mathematics and background of the discussed techniques, you will gain hands-on experience in using them on real data during the course and interpreting the results.

This course covers both classical and modern topics in (un)supervised data analysis and visualisation:

1. Exploratory data analysis (EDA) for unsupervised problems;
2. Machine learning and statistical learning techniques such as Correspondence analysis, Principal Components analysis and Clustering methods;
3. Supervised learning using structural equation modelling and deep learning.
4. Visualisation (throughout the course).

Format
Each week there will be a lecture and a practical.

Effort requirements 
Attendance of at least 80% of the practicals.
 
Students are expected to have knowledge of:

• Introduction to programming in R (ECB2PR)
• Applied data analysis and visualization I for economists (ECB2ADAVE)

Entrance
For UU students only, no exchange students this year.