Bug fix: background color and general increase of readability and robustness of timeseries plot

United4Surveillance / signal-detection-tool

A tool for detection of signals in infectious disease surveillance data.

Other

8 stars 0 forks source link

Bug fix: background color and general increase of readability and robustness of timeseries plot #259

Closed jaemol closed 7 months ago

jaemol commented 7 months ago

Resolves #246

In this PR bug fix:

Correct missplaced bracket, leading to error for some countries.
Fix blue background colouring for when using facet for strata timeseries in report
Only show every other month for report timeseries of strata using facet -- Would like input on this -- Does this increase readability or not necessary?
Make y-axis depend on facet max value (scales = "free-y"), rather than a single ymax value. This is done, since some facet plots could be very difficult to read. -- Would we like this? It demands the user to notice the difference y-axis values.

Input welcome. :)

tinneuro commented 7 months ago

Thanks for your great work and ideas additional to what was required in the issue. I somehow was blind to the bracket problem with the max_date now I see it. :) I think displaying only every two months definetly increases readability. Another option in general for our timeseries visualisation is to show years and months instead, something similar to this maybe: showing_dates But then not sure where to put the month as we have several datapoints for each month. So an alternative would be also this: showing_dates2 thus we could get rid of the day number and just write the calendarweek which also increases readability.

I like that you added the "free_y" to the facet_wrap() - but we have too many y values being displayed and overlapping. Thus we would need some setting that from min and max value we put x many values or so.

overlapping_y_axis_labels

But we can just make a new issue for my suggestions as well and ask someone from the Austrian team to work on that :)

jaemol commented 7 months ago

Thank you for your input!

I think it would be great to make a separate issue regarding the date x-axis. :)

In regards to the free_y, I here propose to just reduce the number of pretty break. This makes it a lot easier to read, and as far as I can tell, it did not mess up the plotly::ggplotly version of the timeseries, neither in-app or in report.

What do you think?

tinneuro commented 7 months ago

I think it looks good, great! Agree on an additional issue for the x axis in the report.