Institutional Repository of Academic Research University of Macedonia
Not a member yet
2215 research outputs found
Sort by
Predicting the relative performance among financial assets: A comparative analysis of different approaches
No full textWe perform a comparative analysis of a wide array of approaches for the problem of forecasting the relative performance among different tradable assets in the framework of the M6 competition. To produce the forecasts, we employ various models spanning probabilistic, classification, and time-series methods, each approaching the problem from a different perspective. We demonstrate that in the case of financial forecasting, simple machine learning approaches have better performance compared to more complex deep-learning models. Furthermore, approaching the problem as a classification task appears to be beneficial. We also confirm findings from existing literature that using simple ensemble techniques can improve performance, and that forecasting performance is better for exchange-traded funds and assets that have lower idiosyncratic volatility. Finally, we benchmark our results against the performance of teams that participated in the M6 competition
The generosity ecosystem framework (GEF): a comprehensive approach to giving
No full textGenerosity plays a crucial role in addressing economic and social inequalities worldwide. Generosity research has garnered attention from various disciplines; however, to date, it has largely taken a piecemeal approach, focusing on either the generous individual or the recipient. Aiming to fill this gap, we conducted an extensive literature review on generosity and developed the proposed Generosity Ecosystem Framework (GEF). GEF analyzes the interactions and influences of various actors and their outcomes at different ecosystem levels (micro, meso, and macro). The interchange between the micro, meso, and macro levels creates a dynamic ecosystem where generosity is cultivated. Individual actions inspire community engagement, and vice versa, whereas organizational collaborations amplify these efforts within a supportive societal framework. Understanding these interrelations is vital for developing effective strategies to promote generosity at all ecosystem levels. GEF contributes to the nonprofit marketing, service, and business literature by reducing conceptual ambiguity, providing a theoretical foundation for future research, and enabling actionable decisions.22348951
An Effective Investments System in the Optimal Portfolio Selection Intelligence (OPSI)
No full textThe optimal portfolio selection problem is investigated in fundamentals of higher order moments. The returns behavior frequently skewed and in excess kurtosis, along with investors’ preferences set new grounds of discussion. Higher order moments, than the kurtosis, will offer further information on investors. A more complex problem arises, of higher flexibility, non-convexity, in unlimited scale fitted to portfolio optimization. The principal problem of Free Will is thus answered, with emphasis on investors. We discuss the OPSI model introducing three hybrid neuro-genetic models of numerous topologies and one regression. Firstly the Radial Basis Function Networks-RBF are in 40 hybrid forms and 10 RBF Neural Nets whilst the results are compared to 50 Time-Lag Recurrent Network-TLRN Hybrids topologies, 10 on the MultiLayer Perceptron-MLP Neural Nets, and the Bayesian Logistic Regression-BLR, to define the most competitive methods in asset allocation and corporate evaluation. New solutions are offered under specific hybrids whilst portfolio efficiency is either evolutionary or intelligent. Introducing the parameters of financial health, we propose the advanced expected utility function filtering noise. The problem of wealth maximisation is transformed to a preferential combination on gain and loss. The TLRN hybrid networks are a very efficient and reliable model on portfolio selection. The OPSI model offers a competitive approach in efficient portfolio selection, protecting the investor from systematic exposure. In the investors Free Will problem, the answer is that Logic is dynamic linearly but adjusting to the environment overrides new challenges of superior potentials than the linear series of events. It is consistent to the maximisation of utility and investors’ welfare.13
Technical and environmental inefficiency measurement in agriculture using a flexible by-production stochastic frontier model
No full textIn light of the urgent need for farms to mitigate greenhouse gas emissions while maintaining economic viability, this paper analyses technical and environmental inefficiencies and their determinants based on a flexible multi-equation by-production stochastic frontier model, which accounts for the stochastic dependence between good and bad outputs via a copula function. An empirical application to Dutch dairy farms illustrates the distortions in the inefficiency scores and in the estimates of their determinants that occur when the dependence between good and bad outputs is ignored. The empirical results indicate a strong positive dependence between the good output (milk) and the bad output (methane emissions), which is particularly pronounced in the upper tail of the distribution. Notably, farms exhibit high efficiency in maximising their good output and minimising their bad output. Subsidies are negatively related to good output inefficiency but positively related to bad output inefficiency, while stock density exhibits a negative association with inefficiencies in both outputs. Disregarding output dependence leads to distortions in inefficiency estimates, particularly affecting the estimates for their determinants.76116418
A copula-based semiparametric by-production stochastic frontier model
No full textThis paper introduces a flexible semiparametric approach to by-production stochastic frontier analysis. We model a firm's production process using two interrelated technologies: one for generating good output and another for generating bad output, both driven by a common set of inputs. Technical inefficiency is defined as the failure to maximize good output, while environmental inefficiency reflects the failure to minimize bad output. Unlike previous studies, our method employs splines to capture nonlinear relationships between inputs and outputs, as well as between firm characteristics and inefficiencies, while also incorporating random firm effects. We model output dependence using copulas and compare alternative specifications. Applying the framework to Dutch dairy farming, where milk is the good output and methane emissions the bad, we find evidence of nonlinearities, low inefficiency levels, and moderate positive dependence between outputs. Neglecting nonlinearities and random effects inflates inefficiency estimates, highlighting the importance of flexible modeling in future applications.15210724
Fuel price effects on motor vehicle collisions: Evidence from Greece
No full textThis study examines the relationship between petrol prices and vehicle collisions using Greek data from 2012 to 2021. Generalized autoregressive conditional heteroscedasticity models are employed for daily motor vehicle collisions. Our analysis reveals that petrol prices have a significant impact on vehicle collisions. Fatal vehicle collisions decrease during relatively high petrol prices, whereas light-injury vehicle collisions increase. No significant relationship was found between severe-injury vehicle collisions and fuel prices. We also analyze daily data on motorcycle vehicle collisions and find a positive relationship between these accidents and fuel prices. When considering models with lagged fuel prices, our results indicate that in all cases, vehicle collisions decrease during periods of increasing fuel prices. These findings suggest that policies targeting motorcycling safety are particularly necessary during times of rising fuel prices.32e0043
Bridging FANETs and MANETs for synchronous data collection in precision agriculture activities using AirPro-FL: An energy aware fuzzy logic routing protocol
No full textThe use of Flying Ad-hoc Networks (FANETs) in precision agriculture requires the development of advanced routing protocols to manage UAV-specific challenges effectively. This paper presents AirPro-FL, a proactive routing protocol that uses fuzzy logic to optimize UAV performance in precision agriculture tasks. Unlike conventional FANET research, which often relies on stochastic mobility models that do not accurately reflect real-world agricultural missions, AirPro-FL is designed to address these gaps by enhancing UAV cooperation in scanning operations such as crop scouting, crop surveying and mapping, spraying applications, and geofencing. Traditionally, these agricultural activities rely on a single UAV, often resulting in inefficiencies. The UAV's limited real-time data transmission capabilities, vulnerability to operational failures, and potential mission execution delays contribute to reduced overall effectiveness. The proposed system involving multiple UAVs significantly speeds up mission completion and enables real-time data transfer through the cooperation between FANETs and Mobile Ad-hoc Networks (MANETs). This innovation empowers agricultural stakeholders to make faster and more reliable decisions based on accurate data collection. Simulation results indicate that AirPro-FL consistently achieves the highest Packet Delivery Ratio (PDR) across all scenarios, halves the average end-to-end delay compared to the second-best protocol, and exhibits superior energy efficiency. The protocol's success in optimizing data collection during scanning operations underscores its broader applicability beyond agriculture, extending to other fields such as environmental monitoring, disaster management, and surveillance, where similar mobility patterns are employed.3110153
The Green Road to Open Access: Policy, Incentives, Actions
No full textThe Library of the University of Macedonia, through the institutional repository of
academic research RuomoPlus, has adopted and actively implements the
University’s Open Access (OA) policy, primarily supporting the Green Open
Access route. Within this strategic framework, researchers are actively
encouraged to deposit the full text of their research publications in the
Institutional Repository, ensuring the preservation, dissemination, and long-term
accessibility of the University's scientific output.
Since 2008, the University of Macedonia has established a procedure for
awarding Research Awards to articles published by university researchers in
high-impact academic journals. The regulation governing these awards was
amended five years ago, introducing a new requirement: for a publication to be
eligible for evaluation and award, all of the candidate researcher's publications
31o Πανελλήνιο Συνέδριο Ακαδημαϊκών Βιβλιοθηκών – 22-24/10/2025 Ιωάννινα
Το κείμενο της ανακοίνωσης (conference paper) είναι διαθέσιμο στο:
https://olympias.lib.uoi.gr/jspui/handle/123456789/39368
for the relevant year must be deposited in the Institutional Repository. This policy
has turned the Research Awards scheme into a strong incentive for academic
staff and researchers to systematically engage in self-archiving.
As a result of this initiative, more than 60% of the University’s publications from
the past five years indexed in the Scopus database are now openly accessible
through the Repository, significantly enhancing the global visibility and impact of
the University's research output.
At the same time, the University of Macedonia Library plays an active and
multifaceted role in promoting OA. It has developed an online information portal,
which serves as a hub for support and training for researchers, librarians, and
other members of the academic community. The portal offers guidance, tools,
and updates on topics such as copyright, repositories, and open science
policies.
This comprehensive approach clearly highlights the role of academic libraries
not only as centers for information management but also as key agents in the
transition toward a more open, participatory, and equitable scholarly
communication ecosystem
Blueprint for Integrating Artificial Intelligence in Educational Programming Environments Without Disrupting Pedagogical Coherence
No full textThis work presents a novel programming environment designed to introduce Greek secondary-level learners to fundamental coding concepts through a specialized educational language. Recognizing the limitations of existing tools, the proposed system combines conventional features such as modern code editing and real-time output, with an embedded intelligent assistant. This assistant offers context-sensitive guidance, pinpointing logical missteps and suggesting refinements. The architecture relies on a segmented design, separating core language interpretation from the modules providing dynamic feedback. By anchoring the AI module in verified educational content, the system delivers targeted support without compromising integrity. Beyond immediate classroom benefits, this approach has broader implications for enhancing algorithmic thinking. Overall, the paper underscores the significant promise of integrating cutting-edge methods into purpose-built educational tools, opening the door to more personalized, efficient, and inclusive learning experiences.755358368Artificial Intelligence Applications and Innovation
From Technical Debt to Business Process Debt: A Framework for Proactive Debt Management in BPM
No full textBusiness Process Management (BPM) is a critical methodology for organizations aiming to enhance operational efficiency and achieve strategic objectives. While BPM’s iterative lifecycle approach ensures continuous improvement, it is prone to shortcuts and compromises that prioritize short-term benefits over long-term sustainability. This tradeoff has been previously examined by the authors and is referred to as Business Process Debt (BPD), akin to Technical Debt (TD) in software engineering. However, unlike TD, which benefits from a well-defined set of indicators to identify its presence, the concept of BPD lacks a corresponding framework or established indicators. This paper addresses this gap by proposing a framework to identify potential BPD specifically during the business process modeling phase. Drawing parallels from software metrics used to define TD indicators, the framework leverages validated business process model metrics correlated with the External Quality Characteristic (EQC) of maintainability. The framework introduces the Total Maintainability Score (TMS), which aggregates these metrics to evaluate the overall maintainability of business process models. By classifying models into low, moderate, or high maintainability levels, and recognizing the inverse relationship between maintainability and debt, the TMS provides a practical indicator of potential BPD. The framework’s applicability is demonstrated through a case study of the European Commission’s New Computerized Transit System (NCTS). The findings highlight its potential to facilitate informed decision-making and proactive BPD management, thereby supporting long-term sustainable BPM initiatives.546 LNBIP139154Decision Support Systems XIV. Human-Centric Group Decision, Negotiation and Decision Support Systems for Societal Transition