Quantitative Issues in Cancer Research Working Seminar

Jill LundellResearch FellowHarvard T.H. Chan School of Public HealthAutomatic Tuning of Supervised Learning Methods with Applications in Cancer Diagnosis Abstract: Supervised learning methods can be used to create powerful classifiers, but many models must be tuned to obtain good results. I will present work I have done on utotuning support vector machines, boosted trees, and elastic net. This work was used to create an R package called EZtune which I am also using to classify tumors based on methylation profiles.