Hannes Buchner, Staburo Managing Director, along with working students Maria Blanco and Tim Müller, will present their applied research project titled “Deriving interpretable thresholds for variable importance in random forests by permutation” at the colloquium of the Statistics Institute at the Ludwig-Maximilians-Universität (LMU) on May 10th at 16:00.
The research project is a collaboration with the Staburo employees Laura Schlieker, Hannes Buchner, Maria Blanco and Tim Müller as well as Armin Ott from Roche Diagnostics. Investigating Random Forests is an active field of interest to researchers and practitioners alike, as they are flexible and well-suited to deal with many challenges present in high-dimensional biological data, while having a proven record of offering good predictive accuracy. The group is working on algorithms that compare variable importance scores, with the goal of identifying all relevant variables that play a role in this prediction. The talk will cover the application of the established Boruta algorithm on real and simulated data sets and compare its performance to a variation of the method proposed by the working group.
The colloquium is an academic gathering that provides an opportunity for students, researchers, and professionals from academia and industry to exchange ideas, present and discuss their latest research findings in the field of statistics and its applications. It is open for both in-person and online participation, further information – including how to join and the project abstract – can be found in the colloquium’s webpage.
We are excited to present our work at the colloquium and extend an invitation to all interested parties to attend and participate in the discussions.
