Forecasting Inflation with Machine Learning (2023)

Undergraduate: Sarah Larino

Faculty Advisor: Andrii Babii
Department: Economics

---

In this research, I employ a number of machine learning (ML) methods on the inflation forecasting problem space. I utilize macroeconomic indicators alongside textual data and apply ML methods to an updated time horizon. Ultimately, I find that ML methods are a viable alternative to traditional benchmarks under certain time horizon conditions, particularly with the inclusion of textual data. However, in contrast with the previous literature, I demonstrate that some ML models are particularly sensitive to the treatment of outliers. When a full time horizon is employed and outliers are included, certain ML models that performed well in previous analyses are not able to outperform other forecasting methods.

Link to Abstract Link to Poster