search

ramhiser / paper-hdrda

High-Dimensional Regularized Discriminant Analysis
0 stars 0 forks source link

Add timing comparison #14

Closed ramhiser closed 8 years ago

ramhiser commented 10 years ago

Use the rda function from the klaR package as baseline

ramhiser commented 10 years ago

Here are comments from the first AoAS reviewer that may be beneficial.

Major Comments

3) The selling point of this paper is the improved computational efficiency as claimed. What is the relative scale of improvement for real data analysis. Is it days vs. hours? Or is it minutes vs. seconds? The author should also discuss the computational complexity, i.e. where the computational speed up took place compared to the original rda.

4) A computationally more efficient implementation of Friedman 1989 was presented. While the work is worth publishing, Annals of applied statis- tics may not be the most suitable. The manuscript might be more suit- able for journals such as statistical computing, computational statistics or a software journal.