A Study on Visual Representations for Active Plant Wall Data Analysis
Round 1
Reviewer 1 Report
The authors report on a user study of three different types of visualization applied to the analysis of data coming from sensors that monitor plant walls. The data is multivariate (i.e. multiple sensors at the same time) and temporal, which makes it complex enough that simple straightforward visualizations are typically not enough to meet the analyst's needs, which then motivates such a study.
# Pros
* It is a novel work in an under-explored knowledge domain.
* The paper reports on a user study, which means it is useful for researchers who wish to build new visualizations in this domain.
* The study of multivariate temporal data in different domains is timely and relevant.
# Cons
* I feel like the paper is too long. Section 7 explains in a too much details the conclusions of each of the different parts of the study (overview, details, etc.) when it could have been a matter of quickly pointing the order of the visualizations (from most to least useful, for ex.) and some examples of reasons. It also feels like the subsections on discussions simply repeat what was already stated in the previous subsection (user feedback), so maybe it is not necessary to have so many subsections.
* Although the study is focused on active plant walls, there is very little advice, discussions, or lessons learned for researches on this domain (or similar domains). The studied visualizations are generic and could be applied to any type of temporal multivariate data, correspond to very common types, and the results of the study basically agree with previous work, which means the contribution is not in the visualization side. If the contribution is in the design of data-driven interfaces for active plant walls, then the authors need to relate more to this target audience.
* The analysis of user feedback is too generic and specific excerpts are only discussed in passing. Since this is a report on a qualitative study, I believe the qualitative results must be more detailed. Also, the results must be more directed towards the specific domain experts, which are the target audience and must have a better motivation for reading the paper.
# Conclusion
Right now I feel like the paper could still be accepted if the three points commented above are treated in a satisyfing way, so I recommend a major revision.
As an advice for the future, I'd say that although the idea of not having interactions for keeping the experiment simple is indeed reasonable, I feel like it was slightly misused in this case. For example: some of the users gave feedback on the fact that analysis was easier if the views of two variables were close to each other; that means some pairs of variables which were present in the tasks were far away. That certainly introduced bias in the results, since we can see for example from Figure 9 that each visualization arranges itself in different ways: either in a grid, or in a shared space, or in split rows. It seems to me like the experiment results must be extended with the lessons learned from this experiment, so that in a future study the analysis is more naturally tied to the required tasks.
# Examples of Typos
* Line 107: Fig. Fig.
* Line 145: domain expert(s)
* Line 309: reve(a)ling
Author Response
Dear reviewer
We first would like to thank you for the thorough feedback on our submission. We have used the comments from the review process to adjust our manuscript. Below we have submitted our replies in a pdf file.
On behalf of all authors,
Kahin Akram Hassan
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors analyzed the effectiveness of three data visualization approaches (i.e., line graph, stacked area graph, and horizon graph) in communicating the status of active plant walls. Five domain experts in the related field were selected and interviewed to help qualitatively evaluate their effectiveness. However, it is not clear why the active plant walls were selected as the study field and how the qualitative results from this study would contribute to the monitoring and visualization of the health status of an active plant wall? The three data representation methods (line graph, stacked area graph, and horizon graph) used in this study are in fact commonly introduced and applied for data representation in many fields; people are very familiar with these basic graph options and all understand that selection of an appropriate graph option is depending on data structure and various application objectives. I don’t see any advances by applying these common graph techniques to representing the monitoring data of active plant walls. Besides, it is not clear why the so-called domain experts should be selected and interviewed for evaluation purposes? Do you suggest that your visualization graphs are difficult to understand for the general public? In most cases, our efforts will be focused on the general public as the domain experts have the right knowledge and expertise to understand the data without visualizing them; however, an average person usually does not have the required knowledge or expertise, thus more user-friendly and easy-to-understand graphs should be used for data representation. In addition, there are many existing or emerging methods for representing high-dimensional data, such as cluster trees, 3D plots, heatmaps, animation, etc. Why don’t you explore these methods in your study? Some other comments are listed as follows:
(1) All figures: it is difficult to understand the effectiveness of these graphs without providing the description of the true variable(s) included in each graph. The authors should at least provide the true variables included in the example dataset. It would be better if the spatial and/or temporal features of each variable could be provided such that the readers can better understand the effectiveness of these graphs in a real context (e.g., for an active plant wall).
(2) How you define the so-called domain experts? Why only five experts were selected in your study?
(3) As all the answers from the expert interviews are all qualitative, how do you systematically score or weight them?
Author Response
Dear reviewer
We first would like to thank you for the thorough feedback on our submission. We have used the comments from the review process to adjust our manuscript. Below we have submitted our replies in a pdf file.
On behalf of all authors,
Kahin Akram Hassan
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have done a lot of work to accommodate the reviewers' comments, which improved the paper significantly. I am happy with the current version and I believe the paper is as good as it can be for now, so I recommend acceptance.
Author Response
p.p1 {margin: 0.0px 0.0px 13.3px 0.0px; line-height: 15.0px; font: 13.3px Arial; color: #000000; -webkit-text-stroke: #000000} p.p2 {margin: 0.0px 0.0px 0.0px 0.0px; font: 12.0px Helvetica; color: #000000; -webkit-text-stroke: #000000} p.p3 {margin: 0.0px 0.0px 12.0px 0.0px; line-height: 15.0px; font: 11.0px Helvetica; color: #000000; -webkit-text-stroke: #000000; min-height: 13.0px} p.p4 {margin: 0.0px 0.0px 0.0px 0.0px; font: 11.0px Helvetica; color: #000000; -webkit-text-stroke: #000000} span.s1 {font-kerning: none}
Dear reviewer
Thank you for all your feedback. With your comments, we believe that the paper has been further improved, and we are thankful for your contribution.
On behalf of all authors,
Kahin Akram Hassan
Reviewer 2 Report
The authors have addressed most of my comments. However, I am still not convinced by their explanation for selecting only a few domain experts in the experiment. Even for qualitative research, you still need to include enough samples to minimize the irreducible biases associated with people's preference, education background, practical experience, etc. Besides, the definition for "domain experts" is not clearly provided in the revised manuscript.
Author Response
Dear reviewer
Thank you for your feedback, with your comments, we believe that the paper has been further improved, and we are thankful for your contribution. We have added our reply in the pdf file below.
On behalf of all authors, Kahin Akram Hassan
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
The authors have addressed my comments properly.
