Democritus University of Thrace (DUTH)
Big data analytics in economics: techniques and applications
Abstract
dc:descriptionAdvancements in applied econometrics have significantly reshaped how economic phenomena are analyzed, forecasted, and evaluated. The increasing availability of high-frequency and granular data, combined with machine learning (ML) techniques, has provided new opportunities to enhance forecasting accuracy, improve causal inference, and incorporate heterogeneity in economic models. This thesis contributes to this evolving landscape by integrating advanced econometric methodologies with ML to improve predictive performance and policy evaluation. Through two case studies in real estate economics and energy economics, this research demonstrates how machine learning-enhanced econometrics can refine estimations, optimize policy interventions, and reduce costs in experimental setups. The first study addresses real estate economics, where precise house price forecasting and bubble detection are critical for financial stability and market efficiency. Traditional econometric models often fail to capture spatial heterogeneity and cross-sectional dependencies, leading to biased estimations. By incorporating heterogeneous panel estimators within a ML framework, this research enhances forecasting models to improve early warning mechanisms for real estate bubbles. This methodological advancement not only benefits financial markets and investors but also provides policymakers with tools to prevent housing-driven economic crises. The second study focuses on energy economics, where causal inference plays a crucial role in designing effective interventions for energy conservation. Behavioral-based policies, such as home energy reports (HERs), are widely used to encourage energy-efficient behavior. However, conventional estimation methods often overlook the substantial heterogeneity in consumer responses. This research integrates causal machine learning techniques to estimate heterogeneous treatment effects (HTEs) and target the most responsive consumer segments. By leveraging these insights, the study optimizes intervention strategies, demonstrating how precision-targeted policies can achieve greater energy savings at a lower cost. Both studies contribute to the broader field of applied econometrics, emphasizing the importance of incorporating heterogeneity in forecasting models and causal inference frameworks. The integration of econometrics with ML enhances the precision of estimations, leading to more efficient policy design and cost reduction in experimental setups. This research highlights the importance of methodological advancements in economics, demonstrating how improved analytical techniques can lead to more precise economic forecasting, effective market regulation, and well-informed public policy. By bridging the gaps between the traditional econometric techniques and the modern machine learning approaches, this thesis provides a framework for improving analytical outcomes in diverse economic settings. Future research should explore further applications of these methodologies in areas such as climate policy, financial risk management, and taxation strategies, ensuring that economic models remain adaptive and responsive to an increasingly complex world.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Karypidis, Paris-Alexandros
- Καρυπίδης, Πάρις-Αλέξανδρος
Subjects
dc:subject × 30- Οικονομία των ακινήτων
- Μηχανική μάθηση
- Πρόβλεψη τιμών
- Φούσκες τιμών
- Ετερογένεια
- Ανάλυση μεγάλων δεδομένων
- Οικονομία της ενέργειας
- Χωρική ετερογένεια
- Διαστρωματικές εξαρτήσεις
- Συμπεριφορικές παρεμβάσεις
- Αιτιώδης μηχανικής μάθησης
- Ετερογενείς επιδράσεις παρέμβασης
- Real estate economics
- Machine learning
- Price forecast
- Price bubbles
- Heterogeneity
- Big data analytics
- Energy economics
- Spatial heterogeneity
- Cross-sectional dependencies
- Behavioral interventions
- Causal machine learning
- Heterogeneous treamtment effects
- Οικονομικά και Επιχειρήσεις
- Κοινωνικές Επιστήμες
- Οικονομικά
- Economics and Business
- Social Sciences
- Ecomonics
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/59624