Update decision tree to include an option for handling masked points #2035
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #2035 +/- ##
==========================================
- Coverage 98.39% 98.34% -0.05%
==========================================
Files 124 132 +8
Lines 12212 12842 +630
==========================================
+ Hits 12016 12630 +614
- Misses 196 212 +16 ☔ View full report in Codecov by Sentry. |
|
|
||
| expected_mask = np.full_like(precip_cube.data, False) | ||
| expected_mask[0, 0] = True | ||
| expected = np.ma.masked_array(expected_array, mask=expected_mask) |
There was a problem hiding this comment.
This is suspicious. The test at the end looks at expected_array and yet this test still passes. This problem is hidden by another problem in the test at the end - see later comment.
There was a problem hiding this comment.
I've updated the decision tree slightly so the if masked will go to false rather than true which would stop this test passing.
| @@ -74,9 +74,12 @@ accessed with this key contains the essentials that make the node function. | |||
| must be exceeded or not exceeded (see threshold_condition) for the node to | |||
| progress to the succeed target. Two values required if condition_combination | |||
| is being used. | |||
| - **threshold_condition** (str): Defines the inequality test to be applied to | |||
| - **threshold_condition** (str or list(str)): Defines the inequality test to be applied to | |||
There was a problem hiding this comment.
I think we should have a unit test that demonstrates this new behaviour explicitly, rather than via the if_masked functionality.
There was a problem hiding this comment.
I've separated the masked data tests into their own test rather than parameterising it which I think is what you were asking for?
Addresses: https://metoffice.atlassian.net/browse/EPPT-1443
This PR updates the categorical decision tree plugin to include the option to add an "if_masked" key to a node. This defines what node masked points should be taken to.
This should not change the default behaviour of the decision tree so there's no need for the IMPROVER weather symbols decision tree to be updated.
Testing: