We will finish our[masked] meetup series with a great event. The meeting will focus on forecasting, prediction and ensemble Learning and speakers will include Prof Grushka-Cockayne and Dr Prasad Patil. We will have food and plenty of time for networking with other R and Bioconductor professionals.
If you're new to the idea of ensemble learning or forecast aggregation, there is a nice primer at https://towardsdatascience.com/simple-guide-for-ensemble-learning-methods-d87cc68705a2
Prof Grushka-Cockayne will speak on forecast aggregation. She is an excellent speaker and we were incredibly impressed by her when she spoke at a recent Women in Data Science meeting. We are sure her experience of developing ensemble prediction algorithms on large-scale, streaming, noisy real-world data will have important lessons for all of us in genomics.
Prof Yael Grushka-Cockayne
Associate Professor of Business Administration at Harvard Business
Associate Professor of Business Administration, Darden School of Business School
Yael Grushka-Cockayne is a visiting Associate Professor of Business Administration at Harvard Business School and an Associate Professor of Business Administration, Darden School of Business. Her research and teaching activities focus on decision analysis, forecasting and estimation, project management, business analytics, and data science.
Early research in forecast aggregation focused on Bayesian approaches that account for forecast accuracy and correlation among forecasts. Other research has examined why in practice such theoretically sound methods are often outperformed by simpler techniques such as the average. In recent years, the wisdom of the crowd literature has encouraged the exploration of alternative heuristics that performed better than even the simple average in some cases. In this talk, I will review alternative methods and consider how they are evaluated. I will provide guidance on when various methods perform well, and how one might use forecast aggregation in practice.
Dr Prasad Patil
Recently completed his postdoc with Prof Giovanni Parmigiani in Harvard TH Chan School of Public Health and did his PhD in Biostatistics from the Johns Hopkins Bloomberg School of Public Health with Jeff Leek.
Prasad will discuss ensemble prediction and discuss his recent study
Patil P, Parmigiani G (2018). Training replicable predictors in multiple studies. Proc of the Natl Acad Sci,[masked]),[masked].
Abstract of that article: This article considers replicability of the performance of predictors across studies. We suggest a general approach to investigating this issue, based on ensembles of prediction models trained on different studies. We quantify how the common practice of training on a single study accounts in part for the observed challenges in replicability of prediction performance. We also investigate whether ensembles of predictors trained on multiple studies can be combined, using unique criteria, to design robust ensemble learners trained upfront to incorporate replicability into different contexts and populations.
Thank you to Paramount Recruitment for sponsoring food. https://www.pararecruit.com/