linc-tw / covid_breakdown Goto Github PK

Not sure whether it's interesting to show this: a network graph showing the links between confirmed cases if there are any.

The reason is that most cases in Taiwan are weakly linked. The biggest cluster is the military fleet.

Also, is there any interesting (not superficial) information from the network? For example, gender or age histogram is a boring plot. It shows nothing different from a heuristic process.

From a cluster network, one can probably calculate some indicators similar to the R-value. This will be interesting, but it's not necessary to draw a network "graph" to do so. There might be a better way to visualize the information.

Pending & to rethink.

I am working on French translation of the site. Many terms are difficult to translate or their translations are not concise enough.

Hello,Your dataset was added to CoronaWhy (https://www.coronawhy.org/) Data Lake on Dataverse as a piece of common COVID-19 dataframe http://datasets.coronawhy.org/dataset.xhtml?persistentId=doi:10.5072/FK2/CPP6SY

Would you be willing to help with maintenance of your dataset in Dataverse, e.g. adding the relevant metadata and keeping the dataset up-to-date? That will help to make the dataset findable and accessible for medical science community.

Make new pages for data source & copyleft.

Data source

Need to include

• original worksheet,
• link to processed csv files (suggested in #9 ), &
• link to raw csv files.

Copyleft

Need to include

• MIT license for codes,
• MIT license for saveSvgAsPng.js, &
• CC0 license for plots.

Graphs

• It could be nice to have an explanation on what is count and what is coefficient (a hoover could be an option)
• Enable to download data as a csv

Readme

• It is important to make the readme appealing. To this purpose, I would :
  . Project title
  . Motivation
  . Include a Gif showing the interface while navigating on it
  . Mention how to install the necessary libraries, if somebody wants to make the project
  . Give some ideas on where does the project come from
  . Credits
  . Contribute + Contributing guidelines
  . License
  ==> Some good tips on readme can be found here:
  https://medium.com/@meakaakka/a-beginners-guide-to-writing-a-kickass-readme-7ac01da88ab3

Possible upgrades

• Access data by API in order to make automatic updates
• Compare to other countries / Make graphs to compare two countries results
• Simple text mining analysis on local Taiwanese news on Covid to compare what is said with the figures. What are the main keywords that are discussed in local news, ...

Some notes are needed to improve understanding of graphics.

For example, in the 1st plot "cases by transmission", we can actually consider green as a sort of danger level & red+green as a transmission efficiency (R_eff). The 4th plot "days needed to identify cases" actually tells how performing a health authority is.

I am still thinking a way to write down these underlying messages. Putting them as footnotes is more user-oriented & easier to access to, but can make the view messy. Putting them as a separate page keeps the aesthetics, but is more difficult to follow. There is a trade-off to take.

It's reported that at several occasions the display will go wrong. For example, the font becomes too big so texts have been cut. It's annoying because I cannot reproduce these problems.

Still need to figure out which browsers caused these. Some screenshots can be useful.

This will be a plot showing the evolution of testing criteria with regard to time. The criteria have been extended for a dozen of times. Sometimes on new discovered clinical symptoms, sometimes on travel history due to the worsen situation in those countries.

The goal is to show that criteria need to be dynamic to be efficient.

The plot will be with two views, togglable by button. One of them will be a classical timeline with key dates & policy change. Another view will be similar to the screenshot below:

"Confirmed Cases by Detection Channel": should be Not announced

The first time I have seen the pyramid disk view, it took me some time to figure out that the disk pyramid was not a risk pyramid but that the height is the time!

Changing the value per default to "Linear" could help figuring out the dimensions more quickly!

There has to be some information about policies that have been made & their dates. I am thinking of 2 possibilities:

1. Make a separate plot (possibly at a new page) of timeline style or calendar style. In this case I might implement all policies into the plot.
2. Superpose the curve of confirmed cases with relevant policies, such as border restrictions & extended testing.

I might make both.