Feature Statistics - open issues

[lanzagar]

After PR #6158 there are some open issues left that should be discussed:

• [x] Compute Mode for numeric variables?
• [x] Show Mode in the widget to make it consistent with the new (and improved) output? Currently the mode is squeezed into the Median column, which would otherwise be empty for categorical variables. But numeric variables could have both...
• [ ] Documentation should be updated
• [x] There are some warnings which could cause issues in the future:

  Orange/statistics/util.py:510: FutureWarning: Unlike other reduction functions (e.g. `skew`, `kurtosis`), the default behavior of `mode` typically preserves the axis it acts along. In SciPy 1.11.0, this behavior will change: the default value of `keepdims` will become False, the `axis` over which the statistic is taken will be eliminated, and the value None will no longer be accepted. Set `keepdims` to True or False to avoid this warning.
  res = scipy.stats.stats.mode(x, axis)

  orange-widget-base/orangewidget/gui.py:2068: UserWarning: decorate OWFeatureStatistics.commit with @gui.deferred and then explicitly call commit.now or commit.deferred.

[wvdvegte]

If it doesn't require a lot of effort, I think computing mode for numeric variables makes sense - especially if they are integers. And even if they're real numbers it makes sense - among them, there are sometimes round numbers like 0 and 1 that appear in a dataset more often than other numbers. This could be useful information.

[janezd]

Showing mode for numeric variables is trivial. How would it look in the output? Current output variable 'mode' is a string variable because it contains values of different variables (which is what #6185 was mostly about). If this same variable also contained mode for numeric variables, they would be strings? Or would it again be a separate column? @lanzagar?

As for the second warning: after introducing decorators for deferred commits, I changed almost all widgets to use them (#5495). I remember skipping this one, but forgot to document the reason. It could have been that I'd like to avoid recommitting output "Statistics" if only "Reduced Data" has changed. I'l look into it.