R-Ladies @ BSC | Tools for Climate Forecasting & AI for Cancer Immunotherapy

Join us for an amazing workshop, co-organized with BarcelonaR, at the Barcelona Supercomputing Centre!

Programme
18:30 -> Introduction of R-Ladies Barcelona and BarcelonaR user group.
18:40 -> "Collaborative R tools for Climate Forecast Analysis by HPC" (An-Chi Ho and Eva Rifà)
19:00 -> "AI in R: PredIG an explainable XGBoost predictor for cancer immunotherapy" (Roc Farriol-Duran)

"Collaborative R tools for Climate Forecast Analysis by HPC" (An-Chi and Eva)
A large amount of data is used in the study of climate. This is why the use of HPC is necessary to carry out the analysis and processing of climate data. In addition, multidisciplinary workflows require the use of shared tools that originate collaboration between scientists and engineers. The talk presents some R packages developed within the Computational Earth Sciences group of the Barcelona Supercomputing Center (BSC-CNS). It also highlights how collaboration between climate scientists and data engineers using R tools can improve our understanding of climate patterns and lead to better predictions. The use of HPC can speed up calculations and enable the handling of large data sets, making it a valuable tool in climate analysis and prediction.

"AI in R: PredIG an explainable XGBoost predictor for cancer immunotherapy" (Roc)
AI in R is (of course) possible. XGBoost is a state-of-the-art algorithm (Kaggle winner, specially working on tabular data) based on gradient boosting and decision trees that is fully implemented in R (R-package: “xgboost”). Here, we will show an implementation on a biomedical research question: how to predict the immune response against a tumor (and its mutations)?. This case will allow us to focus on a binary classification problem with extreme class imbalance (due to the nature of real patient data) and how (re)tuning XGBoost to deal with imbalance can make it work.

Language: English

Requirements: There are no specific requirements.