a941493eeea57ede8214fd77d41806bc-AuthorFeedback.pdf

However, we strongly believe that our paper contains enough valuable con-14 tributions (as described above); moreover our experiments on real-world data sets from multiple cities show that15 our model significantly improves the performance in predicting the fine-grained mappings from coarse-grained16 areal data. This is very helpful for many disciplines including socio-economics [24], epidemiology [25], public17 security[2],publichealth[13],andurbanplanning[34].18 3. Handling observations: We agree with the reviewer in that introduction of nonlinear link functions (as in the19 warped GP) and/or alternative likelihoods might help handle some kinds of observations (e.g., rates). We will clarify it in our proceeding.