Tobias Sing on Data Mining & Machine Learning
Emerging Research Front Commentary, October 2010
|
Article: ROCR: visualizing classifier performance in R
Authors: Sing, T;Sander, O;Beerenwinkel,
N;Lengauer, T |
Tobias Sing talks with ScienceWatch.com and answers a few questions about this month's Emerging Research Front paper in the field of Computer Science.
Why do you think your paper is highly
cited?
This paper describes ROCR, a popular open-source R package developed by us which is freely available.
It allows users to evaluate the predictive performance of machine learning/data mining approaches—a methodology of ever-growing importance in many fields of science and technology.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
"Data mining and machine learning are key technologies for information technology..."
It does not describe a new discovery. Rather, it describes a tool that synthesizes the current state of the art in classifier evaluation methodology, so that it can be easily used by other researchers.
Would you summarize the significance of your paper
in layman's terms?
Today, data mining and machine learning approaches are used in many areas of science, medicine, and technology, with a vast array of different applications ranging from predicting drug resistance in anti-HIV therapy to automatically identifying spam emails. Before such automated prediction methods can be used, their reliability needs to be evaluated carefully. This is what our tool, described in this paper, allows researchers to do.
How did you become involved in this research, and
how would you describe the particular challenges, setbacks, and
successes that you've encountered along the way?
During my Ph.D. work at Prof. Thomas Lengauer’s Computational Biology department at the Max Planck Institute for Informatics, I needed a software tool for evaluating machine learning approaches for my "main" research (which at that time was in computer-assisted anti-HIV therapy). However, there was nothing available like this. Therefore, I sat together with my colleague Oliver Sander, and we jointly developed this tool mainly for ourselves. We were surprised that this tool was adopted by so many other researchers, and that this by-product of our main research has actually become our most-cited paper.
Where do you see your research leading in the
future?
My Ph.D. was about the optimal use of available drugs for anti-HIV therapy. Since then, I have become very interested in directly contributing to drug development, which is my current activity in the Modeling & Simulation Department of Novartis Pharma AG. In the future, I want to continue to contribute to the development of safe and effective drugs using model-based methods.
Do you foresee any social or political
implications for your research?
Data mining and machine learning are key technologies for information
technology, health-care, and many other fields that are profoundly
affecting our lives and society.
Tobias Sing, Ph.D.
Department of Modeling & Simulation
Novartis Pharma AG
Novartis Campus
Basel, Switzerland
KEYWORDS: DRUG-RESISTANCE.