Background

Machine learning (ML) techniques are increasingly being explored for macroeconomic forecasting and nowcasting, offering enhanced predictive accuracy, flexibility in model design, and the ability to handle large datasets. Algorithms such as Random Forests, Artificial Neural Networks (ANNs), and Long Short-Term Memory (LSTM) networks have demonstrated superior performance in capturing complex nonlinear patterns in high-frequency data. It is for this reason that MEFMI will be conducting a workshop on Machine Learning Approaches to Nowcasting.

This workshop will introduce participants to the application of machine learning models in nowcasting GDP and other key macroeconomic indicators. It will combine theory with intensive hands-on training in Python, equipping participants with the skills to design, train, and evaluate machine learning models tailored to macroeconomic data.

Objectives

The objective of the course is to introduce participants to machine learning models—such as Random Forests, ANNs, and LSTMs—for macroeconomic nowcasting. Through hands-on exercises in Python, participants will learn to prepare data, train models, and assess their performance, while also comparing results with traditional econometric methods.

Content

The course will cover the following core topics:

• Introduction to Machine Learning in Macroeconomics

• Data Pre-processing for Time Series Nowcasting

• Random Forests for GDP Nowcasting

• Artificial Neural Networks (ANNs) and Hyperparameter Tuning

• Long Short-Term Memory Networks (LSTM) for Sequential Data

• Model Comparison and Evaluation Metrics (RMSE, MAE, etc.)

Target Group

This course targets mid- to senior-level officials from central banks and ministries of finance who are engaged in macroeconomic modelling and analysis.