Work detail

Machine learning-based approaches to forecasting international trade

Author: Bc. Tomáš Kovařík
Year: 2019 - winter
Leaders: Ing. Vilém Semerák M.A., Ph.D.
Consultants:
Work type: Bachelors
Language: English
Pages: 44
Awards and prizes:
Link: https://is.cuni.cz/webapps/zzp/detail/191197/
Abstract: In this thesis I focus on comparison of gravity model estimated with ordinary
least squares and Poisson pseudo-maximum likelihood with regression techniques
based on machine learning, namely support vector machines, random
forests, and arti_cial neural networks. I discuss the advantages and disadvantages
of these approaches and compare their forecasting accuracy on
exports data. I demonstrate that random forest models and arti_cial neural
networks provide superior forecasting accuracy

Partners

Deloitte

Sponsors

CRIF
McKinsey
Patria Finance