Luis Torgo - Estimation and Experimental Comparison of Predictive Models

Book Giveaway: Attend this seminar for a chance to receive a copy of the book "Data Mining in R: Learning With Case Studies" authored by Dr. Torgo. Only 1 copy is available.

"This seminar will describe some of the main features of the R package performanceEstimation. This package provides a general infra-structure to compare and evaluate any approach (workflow) to solve any predictive task (classification, regression or time series). The main goals of the package are: (i) to be very simple to use for "business as usual"
comparative experiments (e.g. I want to compare algorithms X, Y and Z on these K data sets); and (ii) to be flexible enough to be possible to use it on very specific experimental settings (e.g. I want to try my new fancy approach that involves my special algorithm, some data-preprocessing steps and some special post-processing stages, on a set of tasks, and I want to evaluate it using some hand-crafted evaluation metrics).

We will illustrate several of these usage cases as well as provide examples on how to explore the results of the experiments. Finally, we will also give examples of the implementation in the package of statistical significance tests along the lines described in the now famous JMLR paper by Demsar (2006), namely through the post-hoc Nemenyi and Bonferroni-Dunn tests and also the associated CD diagrams. If there is time we will also illustrate some features of the package that allow the use of parallel computation (either through multiple cores or computer clusters) for very large comparative experiments."

text source: http://www.liaad.up.pt/news-events/seminarios-do-liaad
image source: https://www.crcpress.com/Data-Mining-with-R-Learning-with-Case-Studies-Second-Edition/Torgo/p/book/9781482234893