Advanced Statistical and Machine Learning Techniques

A workshop offered by the Bonn Graduate Center

This workshop offers a comprehensive exploration of advanced statistical and machine learning techniques, blending theoretical insights with practical hands-on applications. Participants will delve into key methodologies, understand their historical context, and apply them to real-world data sets through guided exercises. This workshop is designed for doctoral students who wish to deepen their understanding of advanced statistical techniques and their applications in various fields.

In-person Workshop

Monday, February 10, 2025
9:00 a.m. - 4:00 p.m.

Trainer

Guido Lüchters

Language

English

Location

Department of Geography
Meckenheimer Allee 176, C-Raum*

Target Group

Doctoral students and postdocs

Track(s)

Research

Certificate

8 units are applicable within the Doctorate plus or Careers plus certificate

* The premises are not barrier-free. Please contact the Bonn Graduate Center if you need assistance.

Eine Wissenschaftlerin und ein Wissenschaftler arbeiten hinter einer Glasfassade und mischen Chemikalien mit Großgeräten.

Description

The course will cover four theories.

1. Theory: PCA and Factor Analysis
• Explore the principles of Principal Component Analysis (PCA) and Factor Analysis.
• Discuss historical perspectives, including the controversial use of these techniques in eugenics and critical evaluations by Stephen Jay Gould (1981).
• Understand the limitations of traditional modeling with many predictors, referencing Babyak's insights (2004).

Hands-on: Car Selling Data (PCA)
• Apply PCA to a dataset on car sales to identify underlying factors and reduce dimensionality.

2. Theory: Statistical Learning
• Insights from the seminal work on statistical learning by Tibshirani and Hastie (2014)
Hands-on: Stepwise Regression of Simulated Data
• Conduct stepwise regression on simulated data to understand model selection and validation.

3. Theory: Structural Equation Model (SEM)
• Study the theory and applications of Structural Equation Modeling (SEM), guided by Chuck Huber's work (2018).
• Explore how SEM can be used to model complex relationships between variables.

Hands-on: Math Scores at University and Kindergarten Data
• Apply SEM to analyze data on math scores from university and kindergarten, interpreting the structural relationships.

4. Theory: Machine Learning (AI)

• Dive into machine learning principles and techniques, drawing from Tibshirani and Hastie (2014).
• Understand the role of machine learning in predictive modeling and AI.

Hands-on: LASSO on Simulated Data
• Implement the Least Absolute Shrinkage and Selection Operator (LASSO) on simulated data to practice regularization and variable selection.

Prerequisites:
A thorough understanding of statistics and familiarity with statistical software is a prerequisite in order to participate in the workshop. Prior participation in the workshop “Introduction to Statistics” or “Statistics with R” is strongly recommended.

Content

Develop a solid theoretical foundation in PCA, factor analysis, SEM, and machine learning
Critically assess the historical and ethical implications of statistical methods
Gain practical experience through hands-on exercises with real and simulated data sets
Enhance skills in statistical modelling, variable selection, and predictive analytics