Practice what you preach :-) This year, I now have been #teaching for 10 years at the Heidelberg University. I started there as a teaching assistant in statistics (I did my PhD working on third-party projects, so I did not have teaching obligations, but I always enjoyed it and therefore volunteered to do so). Then I taught the econometrics 101 course, and in recent years, I have given economics students seminars and courses on data science and machine learning.  Last weekend, it was that time again. I gave a seminar on Generative AI and Large Language Models (LLMs) this year. In a seminar in Germany, students write a seminar paper and present it with the aim of practicing academic writing, so I could mainly learn from them. Students presented on topics such as transformers, challenges of training and evaluating LLMs, environmental aspects of Generative AI, or understanding and consuming APIs.  I enjoy teaching because there is so much to learn from students (e.g. on the details of GPU), what interests them, and which ideas they come up with. In particular, if you are mainly working remotely, as I am, the interaction is fruitful and enriching and helps me a lot with my work at DataCamp, where I do not necessarily meet our learners day-to-day. I am always impressed by how good today's student cohorts are at coding and presenting technical topics in these domains. Many thanks in particular to Christian Conrad for making such courses possible and to the students participating and presenting!  Below is a picture at Heidelberg that my mother took a few years ago while visiting us here. #education #datascience

Today, on July 8th, I successfully defended my Master's Thesis entitled "Curriculum Learning with Automatic Relaxation of Loss Function." This final exam marks the completion of my master's program in Artificial Intelligence specialization and concludes an incredible, demanding five-year journey of studying. With the substantial technical knowledge in Data Science area that I've gained, I am excited to apply this expertise in the business sector. Additionally, the work I conducted during my master's thesis project will be published as a research article in the coming months. The future looks bright!

𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝘀𝘁𝗶𝗹𝗹 𝗼𝗽𝗲𝗻 for elective course 𝗛𝗲𝘂𝗿𝗶𝘀𝘁𝗶𝗰 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗠𝗲𝘁𝗵𝗼𝗱𝘀! 𝗗𝗲𝗲𝗽𝗲𝗻 𝘆𝗼𝘂𝗿 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 in the research field of 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻𝘀 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀. 𝗛𝗲𝘂𝗿𝗶𝘀𝘁𝗶𝗰𝘀 form an indispensable tool for everyone working in operations management. Tinbergen Institute’s field courses in Business Data Science make top-notch faculty and expertise accessible to a broader audience. 📌Join the 𝗛𝗲𝘂𝗿𝗶𝘀𝘁𝗶𝗰 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗠𝗲𝘁𝗵𝗼𝗱𝘀 𝗰𝗼𝘂𝗿𝘀𝗲 during the first block of academic year 2024-2025 (September 2 – October 25) at our location in Amsterdam. The course is part of the two-year 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗺𝗮𝘀𝘁𝗲𝗿 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲 at Tinbergen Institute. Important date: deadline for application for block 1 is August 15, 2024 (https://lnkd.in/eyKSfFhQ) 𝗙𝗮𝗰𝘂𝗹𝘁𝘆:  Wout Dullaert, Professor Supply Chain Logistics at VU School of Business and Economics, Vrije Universiteit Amsterdam (VU Amsterdam) and Daniele Vigo University of Bologna (Alma Mater Studiorum – Università di Bologna). 𝗪𝗵𝗼 𝗦𝗵𝗼𝘂𝗹𝗱 𝗔𝗽𝗽𝗹𝘆? Registration for elective courses during academic year 2024-2025 is open to external participants and (external) PhD students. This offers the opportunity to take a single course to build a new skill or deepen knowledge 🚀 𝗔𝗽𝗽𝗹𝘆 𝗵𝗲𝗿𝗲 👉 https://lnkd.in/eyKSfFhQ   𝗧𝗼𝗽𝗶𝗰𝘀 that will be 𝗰𝗼𝘃𝗲𝗿𝗲𝗱 in the 𝗛𝗲𝘂𝗿𝗶𝘀𝘁𝗶𝗰 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗠𝗲𝘁𝗵𝗼𝗱𝘀 course: 💡 Classical heuristics to construct a feasible solution and improvement heuristics based on structured local search  💡 Metaheuristics aiming at escaping local optima  💡 Including learning methods in metaheuristic search and designing computational experiments  💡 All methods will be illustrated through actual implementations in high-level programming languages. 𝗖𝗵𝗲𝗰𝗸 𝗼𝘂𝗿 𝘄𝗲𝗯𝘀𝗶𝘁𝗲: https://lnkd.in/ehVR_MDZ 𝗔𝗿𝗲 𝘆𝗼𝘂 𝗹𝗼𝗼𝗸𝗶𝗻𝗴 𝗳𝗼𝗿 𝗮𝗻𝗼𝘁𝗵𝗲𝗿 𝘁𝗼𝗽𝗶𝗰? 𝗦𝘁𝗮𝗿𝘁 𝗲𝘅𝗽𝗹𝗼𝗿𝗶𝗻𝗴 𝗮𝗹𝗹 𝗼𝘂𝗿 𝗰𝗼𝘂𝗿𝘀𝗲𝘀: https://lnkd.in/eyKSfFhQ #researchmaster #businessdatascience #datascience #electivecourse #shortcourse #heuristics #OperationsAnalytics

Trying out AI generated summaries of new papers: 🌟 New Paper on Bayesian Causal Forests! 🌟 I'm excited to share our latest research published in the Journal of the Royal Statistical Society Series A: "Bayesian Causal Forests for Multivariate Outcomes: Application to Irish Data from an International Large-Scale Education Assessment.". It's joint work with Nathan McJames, Ann O'Shea, and Yong Chen Goh 📝 🔍 What’s it about? Our study introduces a multivariate extension of the Bayesian Causal Forests (BCF) model. This advanced machine learning model estimates causal effects for multiple outcomes simultaneously, leveraging data from the Trends in International Mathematics and Science Study (TIMSS). 📊 Key Findings: - *Positive Effects of Study Desks:* Access to a study desk at home positively impacts student performance in mathematics (Average Treatment Effect (ATE) 95% CI: [−0.50, 10.14]). - *Negative Consequences of Hunger:* Students frequently feeling hungry at school show a significant decline in mathematics (ATE 95% CI: [−8.86, −1.56]) and science achievement (ATE 95% CI: [−10.35, −0.94]). - *Impact of Absenteeism:* Regular absence from school is linked to decreased performance in both mathematics (ATE 95% CI: [−11.88, −2.27]) and science. 🔧 Technical Highlights: Our model extends the BCF framework to handle multiple correlated outcomes, reducing uncertainty and improving the accuracy of causal effect estimates. This is particularly useful in educational settings where multiple factors influence student achievement. 👨🔬 Why It Matters: Understanding these causal relationships is crucial for designing effective educational policies and interventions. Our findings highlight the importance of providing adequate study environments and addressing issues like student hunger and absenteeism. 📚 Read the Full Paper: For those interested in the detailed methodology and results, the full paper is available here - https://lnkd.in/eekE6_hF. 🔗 DOI:** 10.1093/jrsssa/qnae049 - https://lnkd.in/eekE6_hF Feel free to reach out if you have any questions or would like to discuss the implications of our findings further! #Research #Education #MachineLearning #CausalInference #BayesianStatistics #TIMSS #EducationPolicy --- McJames, N., O’Shea, A., Goh, Y.C., & Parnell, A. (2024). Bayesian causal forests for multivariate outcomes: Application to Irish data from an international large-scale education assessment. *Journal of the Royal Statistical Society: Series A (Statistics in Society)*. https://lnkd.in/eekE6_hF

I wanted to take a moment to reflect on the 5 years I have spent teaching the online Data Science MS capstone course at Rochester Institute of Technology since I am taking a break from teaching. Michael Yacci asked me to teach this course to bring real life data science experience to the students. I would like to thank him for the opportunity. Making an asynchronous online course engaging is challenging. Some students engage fully and others ignore the discussions to the detriment of their final grade. One of my goals was to help students understand the impacts of ethics and bias. "Models are just math,” as someone once told me, but they rest on data that can be filled with bias. Generative AI makes this even more of a concern and data scientists need to be aware of the tools and frameworks to mitigate the harm that can be done. Hopefully the students learned not just how to do a data science project but how they can make a positive impact in an organization. The students in the class came from many backgrounds. There were college professors changing careers. A former foreign journalist now leading a data science team. People re-entering the workforce after an absence for various reasons. Working professionals looking to get ahead. The variety of backgrounds made the discussions all the more interesting. There have been some amazing projects. The navy test pilot triangulating radar sources from in-flight data (the final report was redacted as it was being presented to his chain of command). The cancer researcher working on personalized treatments based on genetic sequencing, research that was published. The work supporting re-manufacturing imaging parts on conveyors to identify them to lessen manual sorting.  Other projects have been fun like the one analyzing the stature and success of hockey goalies. These are just a few of the memorable projects. I wish all my former students the best. Lastly, one bit of advice to learners:  read the syllabus. #datascience #AI #RIT #teaching (Image courtesy of the RIT Brand portal: https://lnkd.in/emVjxmhn)

Algorithms, Data, and Curriculum Studies A thought provoking edited issue by John Weaver and Peter Appelbaum with the Journal of the American Association for the Advancement of Curriculum Studies (JAAACS). Check the issue out here: https://lnkd.in/gGktmTTS Current Issue Vol. 16 No. 1 (2024): Algorithms, Data, and Curriculum Studies Published: 2024-02-21 Editor's Note • Data Science, Algorithms, and Curriculum Studies Articles • An Unsettling Artificial Intelligence: Algorithms, Curriculum, and Futurities (Patrick Phillips, Nicholas Ng-A-Fook) • AI, digitalization, and the emergence of man3: From Enfleshment to disembodiment? (Liang Wang, Bernadette Baker) • Counting Publications: The Curricula of Data and Tenure (Boni Wozolek) • Reading Koopman’s (2019) “How we became our data” as an invitation to resist the formatting of the “informational person” with the support of mathematics education (Eva Jablonka) • A vast literary whiteness and becoming undisciplined: A review of Redlining Culture (Rob Helfenbein ) • Stochastic Parrots, Policies, Octopi Who Pretend to Be Human, & Dancing Robots (Peter Appelbaum) • Bernard Stiegler Philosopher of Algorithms and data science in the AnthropoceneA time to care and dream (John Weaver)

These folks are math students at UFSC, the university where I proudly earned my bachelor's and master’s degrees. These are brilliant students—although they will often underestimate their own capacities, as I've done myself. However, what struck me most is their eagerness to make meaningful contributions to society. While some envision themselves as educators, others aspire to explore an academic research career—traditional paths for mathematicians. But what I tried to help them realize is the vast potential they hold as **decision scientists**. As organizations embark on their digital transformation journey, business leaders are realizing that AI encompasses more than just machine learning. Equally essential are professionals who can construct simulations, build optimization models, and craft algorithms to support better enterprise decision-making. At its core, **AI revolves around mathematical models and algorithms**, with the specific type varying based on the kind of decision that needs to be made. Aside from programming and communication skills—which mathematicians can undoubtedly sharpen—mathematical modeling and algorithm development require a profound **aptitude for abstraction**, which is hard to obtain, but mathematicians have in abundance! Natã Machado and Edson Cilos are unveiling these new horizons to their students and deserve a round of applause! 👏 It was truly a pleasure to share insights and ideas with such a motivated and enthusiastic group. I'm looking forward to interacting more with you all. It was also a blessing to meet Prof. Eliezer Batista again after 10 years and recall some of his favorite quotes, such as "força, coragem e determinação" (strength, courage, and determination). 💪 #UFSC #Mathematics #DecisionScience #DigitalTransformation #ai #Education #Gratitude 🎓🌟

Back to the classroom at Saïd Business School, University of Oxford for Module 2 of the AI for Business Executive Postgraduate Diploma program. On Day 1, Profs Bige Kahraman Alper and Agni Orfanoudaki discussed the intersection of machine learning (ML) and econometrics, common pitfalls in ML, and practical applications of decision trees, among many topics. One key takeaway was the difference between ML and Econometrics: Prediction vs. Causality. While ML excels at making predictions from vast datasets, it struggles to establish causality. Econometrics, on the other hand, can identify causal relationships and conduct counterfactual analysis, crucial for robust decision-making. Understanding the strengths and limitations of each approach is essential for applying the right tool to the right problem. The session also highlighted the power of decision trees as a tool for capturing non-linear relationships in data. They are particularly useful in applications requiring transparency and interpretability. However, challenges like overfitting and stability issues can limit their effectiveness. Techniques like random forests and boosting help mitigate these problems. These insights show the importance of combining ML's predictive power with econometrics' causal insights and the strategic use of decision trees in business applications. Looking forward to diving deeper into these topics and more in the coming days!

Enjoy it ! MIT University offers free online courses (no payment required) Here are 10 FREE courses you don't want to miss: 1. Becoming an Entrepreneur? Get key business skills and an entrepreneurial mindset with MIT Launch. 🦾 https://lnkd.in/dA4xpktF 2. Mathematical Methods for Quantitative Finance Learn key math for financial engineering: linear algebra, optimization, probability, statistics, and R. https://lnkd.in/dp4S4ipJ 3. Foundations of Modern Finance I A rigorous, data-driven financial markets framework from MIT professors. 🦾 https://lnkd.in/dg85zzHg? 4. Supply Chain Analytics Master core supply chain methods like statistics, regression, optimization, and probability. 🦾 https://lnkd.in/dza8Z52Q 5. Machine Learning with Python Machine learning basics with Python. MITx MicroMasters in Statistics and Data Science. 🦾 https://lnkd.in/dpx8Mdjw 6. Humanitarian Logistics Discover how humanitarian logistics provides vital aid to save lives and uphold dignity. 🦾 https://lnkd.in/dMuF8UAQ 7. Introduction to Computational Thinking and Data Science 6.00.2x is an introduction to using computation to understand real-world phenomena. 🦾 https://lnkd.in/dTYSex76 8. Understanding the World Through Data Become a data explorer – learn how to leverage data and basic machine learning algorithms to understand the world. 🦾 https://lnkd.in/dPVcztuk 9. Fundamentals of Statistics Develop a deep understanding of the principles that underpin statistical inference: estimation, hypothesis testing and prediction. 🦾 https://lnkd.in/dR6UjaHP 10. Supply Chain Design Learn how to design and optimize the physical, financial, and information flows of a supply chain to enhance business performance 🦾 https://lnkd.in/d-MTJ_5R

Are privacy, security, and commercial constraints on the sharing of data hampering your research in economics and social science? Recent advances in Machine Learning might make the sharing of algorithms an attractive new option. Methods such as Federated Learning (FL), Decentralized Machine Learning (DML), and so on, all allow users to build empirical models without sharing data with each other. If you want to learn from leading experts and network with peers, then you should join the Workshop on Machine Learning for Data & Algorithm-sharing in Social Science Research, organized by the IMT School for Advanced Studies Lucca. The workshop will take place on May 27-28, 2024, at the IMT campus in Lucca, Italy. The workshop will feature keynote speeches and panel discussions, along with a special proof of concept session. By offering a showcase for use cases of algorithm-sharing applications, the workshop will cover topics such as: - Which are the ideal applications of data and algorithm-sharing, and when is algorithm-sharing preferable to data-sharing? - How and when can data and algorithm sharing facilitate the collaboration between stakeholders? - Which is the state-of-the-art of adopted algorithm-sharing solutions in public and private domains?  The workshop is open to researchers, students, practitioners, and policymakers. You can find more information at the workshop website: https://lnkd.in/dyA8PsQv. Don’t miss this opportunity to learn, share, and connect with a community of experts. Register now and spread the word! #MLDAS #IMTLucca #DataSharing #AlgorithmSharing #Economics #SocialScience #MachineLearning