NeurIPS 2020

Coresets for Regressions with Panel Data


Meta Review

This paper provides a coreset method for a generalization of ordinary least squares regression. The reviewers seem to be in agreement that, especially when considering work in the supplement, the approach is interesting and has the potential for impact due to the usefulness of the least squares generalization but also by enabling yet further future extensions. The authors seem to have already agreed to a number of useful changes for the camera ready, and I will just emphasize a couple here: * Moving up GLSE_k to the main text (and discussing novelty) * Incorporating new experiments I also want to point to the detailed list of conditions in R1's updated review (post-author-response). The authors should be careful to read these and address them as best they can.