Patch Rework (Issue #14)

[stefan-grafberger]

Issue #, if available: #14

Description of changes:

• Reworked all analyses to use the new patch abstraction. This should make analyses easier to implement and will also make multi-query optimization easier by allowing us to work on diffs (the patches) instead of only sets of pipelines.
• Patches can now mark if they require later nodes to be recomputed or not. This is in preparation for future optimizations.
• Because the DataCorruption analysis uses the DataPatch Abstraction now, it will be a bit slower due to not having projection push-up built-in. However, this is one of the first optimizations on the to-do list.

Overview of different PatchTypes that we have now:

• Patch
  • DataPatch
    • DataFiltering
    • DataProjection
    • DataTransformer
  • PipelinePatch
    • OperatorReplacement
    • OperatorRemoval
    • AppendNodeAfterOperator
  • ModelPatch

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[stefan-grafberger]

Codecov Report

Merging #15 (24ec3e0) into main (f4fac3a) will decrease coverage by 1.22%. The diff coverage is 97.20%.

@@            Coverage Diff             @@
##             main      #15      +/-   ##
==========================================
- Coverage   95.31%   94.08%   -1.23%     
==========================================
  Files          27       29       +2     
  Lines        3201     3367     +166     
==========================================
+ Hits         3051     3168     +117     
- Misses        150      199      +49     

Impacted Files	Coverage Δ
mlwhatif/execution/_patches.py	93.44% <93.44%> (ø)
mlwhatif/analysis/_analysis_utils.py	63.11% <93.75%> (-33.64%)	:arrow_down:
mlwhatif/analysis/_data_cleaning.py	95.23% <100.00%> (-0.34%)	:arrow_down:
mlwhatif/analysis/_data_corruption.py	98.09% <100.00%> (+0.15%)	:arrow_up:
mlwhatif/analysis/_operator_fairness.py	97.26% <100.00%> (-0.08%)	:arrow_down:
mlwhatif/analysis/_patch_creation.py	100.00% <100.00%> (ø)
mlwhatif/analysis/_what_if_analysis.py	80.00% <100.00%> (+1.05%)	:arrow_up:
mlwhatif/execution/_multi_query_optimizer.py	100.00% <100.00%> (ø)
mlwhatif/instrumentation/_dag_node.py	100.00% <100.00%> (ø)
mlwhatif/instrumentation/_pipeline_executor.py	99.48% <100.00%> (-0.03%)	:arrow_down:
... and 1 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.