Angolla Praveen Goud

This project studies the two-way relationship between brewery activity and local economic conditions across different regions in the United States. The goal is to understand whether changes in brewery activity affect the economy, and whether changes in the economy also influence the growth of breweries. To complete this study, brewery data was combined with regional economic data such as employment, income, and economic output. The data was organized over multiple time periods to observe changes over time. A cross-lagged panel modeling approach was used to examine how earlier changes in brewery activity relate to later economic outcomes, and how earlier economic conditions relate to later changes in brewery activity. Basic spatial information was also included to account for differences across regions. The results show that brewery activity and economic conditions influence each other. Areas with increasing brewery activity tended to see later improvements in employment and income. At the same time, areas with stronger economic conditions were more likely to experience growth in the number of breweries in later periods. This suggests that breweries can both contribute to local economic growth and benefit from strong local economies. Overall, this project shows that there is a feedback relationship between brewery activity and the economy. The findings help explain how small and medium-sized industries like breweries can play a role in local economic development, while also depending on economic conditions for their own growth.

Globalization has modified supply chains by introducing a high level of interconnections of supply chains across economies. Consequently, the supply chains of businesses are exposed to uncertainties from different markets; which, in a competitive global market, exposes businesses to the risks of low profit margins. Predictability of supply chains offers businesses a competitive advantage. Trade flows provide insight into supply chain functionality at the global market level and will hence be utilized for the analysis in the project. In order to adequately present a predictability approach for supply chains, both forecasting and identification of abnormal patterns are important. Consequently, two baseline anomaly detection models will first be employed; Isolation Tree and Long Short-Term Memory (LSTM) Auto-encoder. Secondly, three advanced time series forecasting and anomaly detection models will be employed; XGBoost, ST-GCN and GNN Neural Gravity. With the lowest RMSE = 0.1511, the XGBoost V2 was the best anomaly detection model out of the three machine learning models, followed by GNN Neural Gravity at RMSE = 0.1517. Similarly, at R² = 0.9707, the XGBoost V2 performed best in forecasting trades for the US. Using the XGBoost V2 to make forecasts of trade and utilizing all the models to identify future trade anomalies (with XGBoost V2 as the established anomaly detection baseline) provides for an approach for business to preempt potential supply chain disruptions and better strategize for robust supply chains.