GT OMSA Office Hour 3: 2026-02-04 скачать в хорошем качестве

GT OMSA Office Hour 3: 2026-02-04

Meeting Notes by Gemini: Mohammad Ashkani opened the third office hour by inviting questions, leading to SOOMI OH requesting resources for soccer analytics terminology, to which Mohammad Ashkani recommended reviewing previous NYU and UC Irvine Capstone projects, particularly focusing on threat metrics, and agreeing to send those repositories via Discord. Mateusz Faltyn advised a backward learning approach for new domains by starting with provided data and schemas and searching unfamiliar terms, while also clarifying with Nick that the provided soccer data is a starting point, encouraging experienced students to extend it with live feeds. Mohammad Ashkani and Mateusz Faltyn addressed student questions on GitHub workflow, emphasizing the importance of version control for industry work, and confirmed they would share resources for learning Git. For the Bitcoin analytics project, Mohammad Ashkani stressed that the project focuses on finding non-trivial relationships using time series analysis concepts like co-integration and Granger causality, rather than "technical analysis," and emphasized understanding the data generation process and conducting formal Exploratory Data Analysis (EDA) on the provided Poly Market data. Mateusz Faltyn confirmed that the project is an allocation framework validated by backtesting, not a trading framework based on typical chart patterns, and both cautioned against the risk of "fishing for significance" and overfitting. Addressing the scarcity of 360-event data, Mateusz Faltyn and Solomon Flax confirmed it is limited, which Mohammad Ashkani presented as a real-world constraint that requires generating insights from limited resources, as past teams have successfully done. Mohammad Ashkani advised students to approach their projects as products by considering user stories, and promised to provide the outline for the midterm deliverable before the next office hour. Regarding current progress, Mohammad Ashkani and Mateusz Faltyn suggested that students should have a clear sense of direction and be performing EDA, with beginners forming domain-specific questions and experienced individuals focusing on understanding data schemas to build a useful micro-product. Mateusz Faltyn concluded by suggesting that experienced students use the practicum to either master their sponsoring company’s tech stack or learn a completely new one.