Granger output and score

PrimozGodec commented 3 years ago

Issue

Granger causality is missing output with selected rows
Granger causality does not show the p-value score
Line plot does not save settings
Add the input with selected features to the line plot

Description of changes

Granger:

Added output
Added p-value
Changed functionality - to be similar to other widgets button Test start and stop the causality computation
Added auto-apply button
computation with ConcurrentWidgetMixin
Refactoring
Tests

Line plot:

implement settings, context
implement input that select features
Code refactoring
Tests

Includes

[X] Code changes
[X] Tests
[ ] Documentation

ajdapretnar commented 3 years ago

@PrimozGodec I like this very much. I was only a little surprised to see the two selected features appear as two separate plots instead of being plotted as in one plot as two variables.

For this:

I would expect to see this:

Nothing major, just an idea.

Found a bug:

Traceback (most recent call last):
  File "/Users/ajda/orange/orange-canvas-core/orangecanvas/scheme/signalmanager.py", line 936, in __process_next
    if self.__process_next_helper(use_max_active=True):
  File "/Users/ajda/orange/orange-canvas-core/orangecanvas/scheme/signalmanager.py", line 974, in __process_next_helper
    self.process_node(selected_node)
  File "/Users/ajda/orange/orange-canvas-core/orangecanvas/scheme/signalmanager.py", line 605, in process_node
    self.send_to_node(node, signals_in)
  File "/Users/ajda/orange/orange-widget-base/orangewidget/workflow/widgetsscheme.py", line 792, in send_to_node
    self.process_signals_for_widget(node, widget, signals)
  File "/Users/ajda/orange/orange-widget-base/orangewidget/workflow/widgetsscheme.py", line 826, in process_signals_for_widget
    handler(*args)
  File "/Users/ajda/orange/orange3-timeseries/orangecontrib/timeseries/widgets/owlinechart.py", line 574, in set_features
    self.update_plots()
  File "/Users/ajda/orange/orange3-timeseries/orangecontrib/timeseries/widgets/owlinechart.py", line 617, in update_plots
    assert len(self.attrs) == len(self.is_logit)
AssertionError

To reproduce: Yahoo - Granger C. - Line Chart. Select the first couple of variables. Then select just the first one.

PrimozGodec commented 3 years ago

I fixed the error. Hope it would be ok now.

Regarding the suggestion:

Each variable is shown in the separate chart since they may have dependency but the ration of data can be much different between variables. For example, if one is in range 0 - 1 and the second is in the range 0 - 1000 the first one is visible as a line and you cannot see actual concurrence.
It would be actually nice if two lines are selected they are shown in the separate plots. I also think it is technically possible, but I am not sure if I can abuse the AttributeList. It technically can hold a list of lists (which would be required to pass selected lines separately) but I think that it should be just a list of attributes (@janezd).

ajdapretnar commented 3 years ago

Each variable is shown in the separate chart since they may have dependency but the ration of data can be much different between variables.

Good point. We can go with existing solution. It is better than what we had before anyway. :)

biolab / orange3-timeseries