Bee_Colony_Dashboard - Group 9

[manju-abhinandana]

DSCI532-peer-review

---

• Name of the app: Bee_Colony_Dashboard
• Link to the repo in milestone2: https://github.com/UBC-MDS/Bee_Colony_Dashboard
• Link to the proposal.md file where it is the description of the persona and scenario: https://github.com/UBC-MDS/Bee_Colony_Dashboard/blob/main/docs/proposal.md

---

[cuthchow]

Reviewer: Group 8 Persona

• Provide 3 pieces of constructive feedback that could help the app to be better suited to the scenario it was designed for:
  
  1. For a study of the potential causes of colony collapse disorder, a user may want to directly compare single stressors with the overall trend. For instance, there could be a second line overlain on the timeseries plot which plots the various stressors over time.
  2. Further to the above point, the dashboard may provide certain metrics relating stressors and colony collapse. e.g It may indicate the correlation coefficient between a chosen stressor and the number of colonies in a given state/time period.
  3. It may also be interesting to compare the trends in multiple states against on another (ie in the line plot, with multiple coloured lines), rather than only being able to examine a single state at a time.
• Provide 3 pieces of constructive feedback that could help the app to be better suited to the user(persona) it was designed for:
  
  1. Allow for aggregation of time periods in the map filter. The Buzzing Bee users may want to see longer term trends, rather than just the bee loss in a single quarter.
  2. Avoid the mixing of aggregated and non-aggregated data. The user may be confused as to why the map is based on a single quarter time period, whereas the timeseries line plot spans multiple time periods.
  3. Strange to use % for the bee colony stressors in a stacked bar chart, and this makes it harder for the user to compare competing causes for colony collapse. It may be preferable to use a grouped bar chart for each time period, so that direct comparisons between stressors can be made, and the percentage axis makes more logical sense in that case (ie bars wouldn't exceed 100%).

Frontend / layout

• Is the text (plot labels / titles / subtitles) easy to read?
• Are the plots selected appropiated for the data? Can you think of other alternatives or improvements for the plots regarding interactivity/reactivity or plot selection?
• Was the dashboard organized in a way that was easy to read (columns / rows) or you would suggest a change in this sense?
• Would you prefer including tabs / navbars to improve the navigation or you think it is better to visualize everything in one screen by the nature of the content?
• Are the plots or other images described properly for the persona described?
  Yes, the labels and titles are all easy to read, but some of the labels could benefit from having a unit directly applied (ie the numbers on the map could have % symbol to clarify what they represent). Moreover, the y-axis ranges on the line and bar plots should be kept constant, so that when filters are changed, comparisons can be made more easily. As a minor change, the title text should be capitalised. The plot types selected are generally appropriate, although I would suggest using grouped bars instead of stacked bars to depict the colony stressors, as it's harder to compare across stressors when using a stacked bar.
  As for the organisation of the dashboard, one change I would suggest is perhaps keeping the filters on one side and the plots on the other. It may be confusing to some users which filters correspond to which plots, since not all of them are applicable to all of the plots. One such way of organising the plots may be indeed to use different tabs for different plots. Nonetheless, overall the plots and filters are very effective in conveying the relevant information to a hypothetical users from the provided usage scenario.

Backend

• Are the options that you can select in the widgets working fine or have you found some cases where it is not working?
• If there are interactive plots: are the labels clear and easy to understand?
  All of the main options that can be selected in the filters are working fine. The mouseover tooltip interactivity of all the plots is also helpful, as I can get a clear understanding of the numbers behind each data point, rather than just eye-balling the axes.

Documentation

• Was the documentation easy to understand from the README.md how to use the app? How could the documentation be improved?
• Was the app intuitive to use without reading the documentation? What information could be good to include?
  Prior to looking at the README, the app was indeed very intuitive to use. However, I would comment that the fact that the period for the map and the period for the timeseries chart are decoupled may be confusing to some users. The README does help a bit in explaining which filters map to which plots, but again it may be more straightforward to have separate tabs for plots which are for a single time period (e.g map) vs plots over a range (e.g line plot). The README could ideally include installation instructions for users to run the dashboard locally on their machines.

Indicate something you would like to include in your own app after reviewing this one.

• I'm interested by the idea of using two drop down selectors as endpoints for a time interval, as opposed to using a range slider as we did in our dashboard. This approach allow for more precise control over the range, especially in cases where there are many potential time points to choose from.