DTA

Corso Ordinario Secondo Livello

CIALDEA, GABRIELE

etd-09162025-112325

On system identification for economics and finance

Classe Scienze Sociali

SCIENZE ECONOMICHE E MANAGERIALI - SCIENZE ECONOMICHE E MANAGERIALI

tutor Prof. TAMAGNI, FEDERICO
relatore Prof. VANDIN, ANDREA

• Nessuna parola chiave trovata

27/11/2025;

completa

Understanding and modeling complex dynamic systems is a central challenge in economics and finance, where nonlinearity, stochasticity, and limited or noisy data are widespread. While traditional econometric approaches provide valuable tools, methods developed in other disciplines, such as engineering, machine learning, and computational biology, offer other potentially useful perspectives for system identification. In this work, we survey a selection of techniques originating from these fields and discuss their potential applications in economic and financial contexts. We introduce a comparative framework based on key properties such as data requirements, interpretability, adaptability to online settings, and the ability to capture nonlinear and stochastic dynamics. Through this framework, we analyze recent machine learning advances such as Neural Ordinary Differential Equations, Neural SDEs, and Physics-Informed Neural Networks, as well as more classical approaches including Sparse Identification of Nonlinear Dynamics (SINDy), NARMAX models, Volterra series, Variable Length Markov Chains, and Boolean Networks. The analysis is supported by illustrative examples that highlight their applicability to economic and financial problems ranging from policy analysis to volatility modeling.

Nome file	Dimensione
Tesi_Cia...vello.pdf	2.22 Mb
Contatta l'autore