Abstract: This tutorial explores machine learning applications in economics and finance using TensorFlow 2. It starts by examining how TensorFlow and machine learning can be used to solve empirical and theoretical models in economics. It then provides an introduction to deep learning and gradient boosting for structured economic and financial datasets. Next, it discusses how to augment structured datasets with text-based features through the use of natural language processing models. Finally, it examines how generative adversarial networks can be used in simulation and estimation exercises in economics and finance. The code from the tutorial will be provided in a Google Colab notebook.
1. TensorFlow for Economics and Finance
This section introduces TensorFlow 2, a machine learning library that is centered around deep learning and graph-based models. It also explains how methods and models from machine learning can be applied to problems in economics using TensorFlow.
2. Training Machine Learning Models on Structured Datasets
Problems in economics and finance are often solved using linear models and structured datasets. In this section, we will see how performance can be improved by using data in tabular format, but with different models: neural networks and boosted trees. Exercises will center around the use of the Keras and Estimator submodules of TensorFlow.
3. Extracting Novel Features from Unstructured Text Data
While structured datasets are available for many problems in economics and finance, unstructured data remains an under-exploited source of novel features that can improve forecast accuracy and model fit. In this section, we will see how natural language processing can be employed to identify useful features in text data. Exercises will center around central bank communication.
4. Simulating Data with Generative Adversarial Networks
Simulation is often used in economics and finance to generate data that has specific statistical properties or to estimate models. In this section, we will introduce the concept of generative adversarial networks (GANs) and explore their use in economic and financial simulation exercises.
Bio: Isaiah Hull is a senior economist in the research division of Sweden's Central Bank (Sveriges Riksbank). He holds a PhD in economics from Boston College and conducts research on computational economics, machine learning, and quantum computing. He is also the instructor for DataCamp's "Introduction to TensorFlow in Python" course and the author of "Machine Learning for Economics in Finance in TensorFlow 2."