The course introduces key concepts and methods of machine learning for prediction, causal analysis, and data exploration, with an emphasis on when and how these tools are useful in economics and business contexts. The course is application-oriented and uses R throughout.
Participants will learn to:
- Build predictive models (supervised learning):
Estimate and evaluate data-driven models, for example, predicting sales from price, quality, and market characteristics, with focus on out-of-sample performance. - Explore and structure complex data (unsupervised learning):
Identify hidden data patterns such as clusters of similar firms or consumers, and use dimensionality-reduction techniques to summarize and visualize high-dimensional datasets. - Estimate treatment effects (causal machine learning):
Study how interventions and policies affect outcomes (e.g., education on wages), including heterogeneous effects across subgroups, using flexible, data-adaptive methods. - Analyze sequential decision problems (reinforcement learning):
Understand learning and optimal decision-making in dynamic environments with feedback, relevant to areas such as pricing, targeting, and adaptive policy design.
