Report CSV (Issue #5)

[ndgnuh]

Basically this it works, the current output looks like this:

The data used to aggregate is the same as the one used for plotting. Right now there are some code overlap. TODO:

• [x] Generate a pandas.DataFrame to report on the data.
• [ ] Clean up code
  • [ ] Make the gather function collect DataFrame instead of just the np.array (for potential simulation output change)
  • [ ] Make the plot function receive (DataFrame, column,...) or (pd.Series,..)
  • [ ] Rename mis-leading named functions
• [ ] Documentation

[chapuisk]

In fact, there is a big point missing in this report: 100k relative hospitalization, ICU, susceptible and so on.

[ndgnuh]

@chapuisk The number is scaled, the total in the table is not the total at each step, it's the accumulated total in the whole simulation, that's why I use max here. https://github.com/COMOKIT/comokit4py/blob/7762dcb340691b20582140f5b71edbabbf0c9cdc/base/post-processing/comokitreport.py#L66-L71

[chapuisk]

@ndgnuh Ok ! so now we should have each column both relative to 100k and unscaled (side by side or in different table, as u wish). This btw should be done, both here and within, what @RoiArthurB refers as "plot" part of the lib (that should definitely not be labeled like that, but this will be in another upcoming issue)

[RoiArthurB]

his btw should be done, both here and within, what @RoiArthurB refers as "plot" part of the lib

@chapuisk How would you like to plot (graph drawing) single values ? Bar chart ? 🤔🤔🤔

[chapuisk]

his btw should be done, both here and within, what @RoiArthurB refers as "plot" part of the lib

@chapuisk How would you like to plot (graph drawing) single values ? Bar chart ? 🤔🤔🤔

"This" means every simulation outcome should be available as it is now AND relative to 100k "Not labeled like this" means that the objective should be to translate replication into a single csv with all variables, deviation, etc. And then propose to plot things relative to it "New issue" means add external data to plots. More to come on this asap

[ndgnuh]

Since the functionality of generating report should be the same, where regardless of scaling or not, I parameterized the scaling.

The data is scaled to 100k by default. If one wants to get the unscaled version simply add scaled=False to the parameter of generateReport.