NSBE - MA Dissertations
URI permanente para esta coleção:
Navegar
Entradas recentes
A mostrar 1 - 10 de 10628
- Agents at work: utilizing design patterns for AI - agent workflow automationPublication . Veltrup, Leon; Obermeier, Daniel; Batikas, MichailThis thesis evaluates the automation potential of knowledge work tasks using multi-agent AI frameworks. A structured scorecard assesses task suitability across project roles. Selected tasks are implemented with distinct agent design patterns in CrewAI, using GPT-4o. Performance is evaluated via workflow-level metrics, including Task Success Rate and Effective Task Success Rate. Findings reveal that pattern-specific configurations significantly influence agent success, robustness, and resource efficiency. The study provides empirically grounded guidance onorchestrating design patterns for agents with varying task complexity in business analytics workflows.
- Grip intelligence field lab: pet consumption and socio-demographics - linking pet retail activity to public health: a store-level approachPublication . Siebler, Nicolas Gustav; Han, QiweiThis thesis investigates whether pet retail transaction data can predict socioeconomic, demographic, political, and health characteristics at the community level. Using approximately 200 million transactions from 1,526 PetSmart stores across the United States, we examine four research questions: predicting neighborhood urbanicity and housing density, linking pet retail to public health indicators, extracting socioeconomic signals from product assortment semantics, and inferring regional political leaning. Machine learning models achieve R² ranging from 0.42 to 0.89 across targets, with performance depending on whether outcomes reflect behavioral choices or structural determinants. Findings demonstrate that consumption patterns encode meaningful community characteristics, offering novel applications for retail analytics, public health research, and political geography.
- The power of being seen: the impact of Google street view on urban development in the United States - crime & education outcomesPublication . Volpe, Laura; Batikas, MichailThis paper examines whether digital visibility affects urban development by studying the staggered rollout of Google Street View (GSV) across U.S. cities. Treating GSV as an informational shock, the analysis uses staggered Difference-in-Differences estimators to estimate causal effects on urban expansion, housing markets, tourism, population, crime, and education. Results show that GSV has modest but meaningful effects in domains closely tied to perception and information, such as tourism activity, housing prices, and selected crime outcomes, while effects on population and employment are limited. Overall, digital visibility matters, but its impacts are heterogeneous and generally moderate.
- Reconfiguring the short-term rental market: the impact of citi bike station expansions on Airbnb activity and listing survival in New York CityPublication . Sousa, Guilherme; Batikas, Michail
- Seasonal momentum: calendar effects and market eventsPublication . Borges, Guilherme de Lucena Sampaio Ramos; Januário, Afonsocombined into a practical, low-frequency trading strategy for the U.S. equity market. Using daily S&P 500 data from 1975 to 2024, the analysis evaluates the turn-of-the-month effect, preholiday returns, Federal Reserve meeting dates, earnings intensity and seasonal expected returns. Each effect is tested individually and then integrated into a unified strategy. While individual anomalies exhibit modest and time-varying performance, their combination concentrates returns into specific trading windows and improves risk-adjusted performance relative to buy-and-hold, showing the economic importance of predictable timing patterns in equity returns.
- Analyzing rhetorical farming in house of commons of the United Kingdom using large language models: Temporal trends and structural patterns frame typology constructionPublication . Reichert, friederike; Shen, YufeiThis study examines rhetorical framing in House of Commons debates on the United Kingdom's exit (Brexit) from the European Union from 2012 to 2022, using Laqrge Language Models(LLMs) to analyze 62,847 parliamentary speech chunks. Building on Neuman et al.'s (1992) generic frame typology, it investigates how framing strategies have changed over time and varied across parties, referendum positions, and speaker attributes. The computational approach provides scale and precision beyond traditional qualitative methods, showing that Economic and Conflict frames dominated and that shifts in usage tracked key Brexit milestones. The findings demonstrate LLMs' value for large-scale political discourse analysis and clarify how parliamentary actors constructed Brexit's meaning during a period of constitutional turmoil.
- EY & Nova SBE PBL project - data-driven strategic analysis tool for higher education institutions: defining the KPI framework and developing data collection protocolsPublication . Cunha, Afonso; Xufre, PatríciaThis project reports on the development of a business intelligence dashboard, supported by a data pipeline that integrates heterogeneous data from multiple institutional sources. Using only the data that is already available, a set of feasible KPIs was defined. Data from Excel files, PDFs and a Python-generated source were extracted with ChatGPT assistance, cleaned and standardized in Power Query, organized into entity and fact tables, stored in Access and visualized in Power BI, enabling more consistent, data-driven monitoring and decision-making.
- The power of being seen: the impact of Google street view on urban development in the United States - tourism & population outcomesPublication . Goubet, Stephanie; Batikas, MichailThis paper examines whether digital visibility affects urban development by studying the staggered rollout of Google Street View (GSV) across U.S. cities. Treating GSV as an informational shock, the analysis uses staggered Difference-in-Differences estimators to estimate causal effects on urban expansion, housing markets, tourism, population, crime, and education. Results show that GSV has modest but meaningful effects in domains closely tied to perception and information, such as tourism activity, housing prices, and selected crime outcomes, while effects on population and employment are limited. Overall, digital visibility matters, but its impacts are heterogeneous and generally moderate.
- A roadmap to Ai for Ngos: a Cáritas case study - benefits, challenges and ethical considerations of Ai adoptioPublication . Ferreira, Ricardo; Costa, Bernardo ForbesDespite growing interest in artificial intelligence among nonprofits, 95% of AI initiatives fail due to inadequate data foundations. This challenge is particularly acute for federated nonprofit structures, organizations comprising multiple legally autonomous entities under a shared mission, where data fragmentation and inconsistent governance create implementation barriers.This research develops a comprehensive AI enablement framework specifically for federated nonprofits. The framework integrates: (1) a three-tier data governance architecture balancing centralized coordination with local autonomy; (2) a five-level maturity model establishing minimum AI readiness thresholds; and (3) a 24-month implementation roadmap. We adopt established corporate governance principles for nonprofit contexts where hierarchical enforcement is unavailable, emphasizing voluntary coordination and consensus-based decision-making. Validation through Cáritas Portuguesa, a federation of 20 autonomous dioceses, demonstrates that technical infrastructure alone does not ensure AI readiness without governance structures. The research addresses ethical considerations including human-centered design, GDPR compliance, transparency requirements, and multi-stakeholder co-creation for "AI for Good" initiatives. A proof-of-concept Streamlit application demonstrates practical implementation, providing tools for data standardization, KPI monitoring, and AI-assisted insights while maintaining confidentiality and human oversight. This work offers federated nonprofits a practical blueprint for building data infrastructure required for ethical AI implementation while preserving organizational autonomy and mission integrity