Merge pull request #35 from causy-dev/workspace-dataloaders

feat(workspaces): workspace init process (create algorithm, dataloade…

README.md

@@ -62,7 +62,7 @@ This runs a web server on port 8000 which allows you to visualize the causal dis
 Use a default algorithm
 
 ```python
-from causy.algorithms import PC
+from causy.causal_discovery.constraint.algorithms import PC
 from causy.graph_utils import retrieve_edges
 
 model = PC()
@@ -90,61 +90,61 @@ from causy.common_pipeline_steps.exit_conditions import ExitOnNoActions
 from causy.graph_model import graph_model_factory
 from causy.common_pipeline_steps.logic import Loop
 from causy.common_pipeline_steps.calculation import CalculatePearsonCorrelations
-from causy.independence_tests.common import (
-  CorrelationCoefficientTest,
-  PartialCorrelationTest,
-  ExtendedPartialCorrelationTestMatrix,
+from causy.causal_discovery.constraint.independence_tests.common import (
+    CorrelationCoefficientTest,
+    PartialCorrelationTest,
+    ExtendedPartialCorrelationTestMatrix,
 )
-from causy.orientation_rules.pc import (
-  ColliderTest,
-  NonColliderTest,
-  FurtherOrientTripleTest,
-  OrientQuadrupleTest,
-  FurtherOrientQuadrupleTest,
+from causy.causal_discovery.constraint.orientation_rules.pc import (
+    ColliderTest,
+    NonColliderTest,
+    FurtherOrientTripleTest,
+    OrientQuadrupleTest,
+    FurtherOrientQuadrupleTest,
 )
-from causy.interfaces import CausyAlgorithm
-from causy.algorithms.pc import PC_EDGE_TYPES
+from causy.models import Algorithm
+from causy.causal_discovery.constraint.algorithms.pc import PC_EDGE_TYPES
 from causy.graph_utils import retrieve_edges
 
 CustomPC = graph_model_factory(
-  CausyAlgorithm(
-    pipeline_steps=[
-      CalculatePearsonCorrelations(),
-      CorrelationCoefficientTest(threshold=0.05),
-      PartialCorrelationTest(threshold=0.05),
-      ExtendedPartialCorrelationTestMatrix(threshold=0.05),
-      ColliderTest(),
-      Loop(
+    Algorithm(
         pipeline_steps=[
-          NonColliderTest(),
-          FurtherOrientTripleTest(),
-          OrientQuadrupleTest(),
-          FurtherOrientQuadrupleTest(),
+            CalculatePearsonCorrelations(),
+            CorrelationCoefficientTest(threshold=0.05),
+            PartialCorrelationTest(threshold=0.05),
+            ExtendedPartialCorrelationTestMatrix(threshold=0.05),
+            ColliderTest(),
+            Loop(
+                pipeline_steps=[
+                    NonColliderTest(),
+                    FurtherOrientTripleTest(),
+                    OrientQuadrupleTest(),
+                    FurtherOrientQuadrupleTest(),
+                ],
+                exit_condition=ExitOnNoActions(),
+            ),
         ],
-        exit_condition=ExitOnNoActions(),
-      ),
-    ],
-    name="CustomPC",
-    edge_types=PC_EDGE_TYPES,
-  )
+        name="CustomPC",
+        edge_types=PC_EDGE_TYPES,
+    )
 )
 
 model = CustomPC()
 
 model.create_graph_from_data(
-  [
-    {"a": 1, "b": 0.3},
-    {"a": 0.5, "b": 0.2}
-  ]
+    [
+        {"a": 1, "b": 0.3},
+        {"a": 0.5, "b": 0.2}
+    ]
 )
 model.create_all_possible_edges()
 model.execute_pipeline_steps()
 edges = retrieve_edges(model.graph)
 
 for edge in edges:
-  print(
-    f"{edge[0].name} -> {edge[1].name}: {model.graph.edges[edge[0]][edge[1]]}"
-  )
+    print(
+        f"{edge[0].name} -> {edge[1].name}: {model.graph.edges[edge[0]][edge[1]]}"
+    )
 ```
 
 ### Supported algorithms

causy/causal_discovery/__init__.py

@@ -0,0 +1 @@
+from .constraint.algorithms import AVAILABLE_ALGORITHMS

causy/algorithms/pc.py → causy/causal_discovery/constraint/algorithms/pc.py

@@ -12,22 +12,28 @@
 from causy.generators import PairsWithNeighboursGenerator, RandomSampleGenerator
 from causy.graph_model import graph_model_factory
 from causy.common_pipeline_steps.logic import Loop, ApplyActionsTogether
-from causy.independence_tests.common import (
+from causy.causal_discovery.constraint.independence_tests.common import (
     CorrelationCoefficientTest,
     PartialCorrelationTest,
     ExtendedPartialCorrelationTestMatrix,
 )
 from causy.common_pipeline_steps.calculation import (
     CalculatePearsonCorrelations,
 )
-from causy.interfaces import AS_MANY_AS_FIELDS, ComparisonSettings, CausyAlgorithm
-from causy.orientation_rules.pc import (
+from causy.interfaces import AS_MANY_AS_FIELDS
+from causy.models import ComparisonSettings, Algorithm
+from causy.causal_discovery.constraint.orientation_rules.pc import (
     ColliderTest,
     NonColliderTest,
     FurtherOrientTripleTest,
     OrientQuadrupleTest,
     FurtherOrientQuadrupleTest,
 )
+from causy.variables import (
+    FloatVariable,
+    VariableReference,
+    IntegerVariable,
+)
 
 PC_DEFAULT_THRESHOLD = 0.005
 
@@ -55,18 +61,19 @@
 PC_EDGE_TYPES = [DirectedEdge(), UndirectedEdge()]
 
 PC = graph_model_factory(
-    CausyAlgorithm(
+    Algorithm(
         pipeline_steps=[
             CalculatePearsonCorrelations(display_name="Calculate Pearson Correlations"),
             CorrelationCoefficientTest(
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 display_name="Correlation Coefficient Test",
             ),
             PartialCorrelationTest(
-                threshold=PC_DEFAULT_THRESHOLD, display_name="Partial Correlation Test"
+                threshold=VariableReference(name="threshold"),
+                display_name="Partial Correlation Test",
             ),
             ExtendedPartialCorrelationTestMatrix(
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 display_name="Extended Partial Correlation Test Matrix",
             ),
             *PC_ORIENTATION_RULES,
@@ -77,19 +84,24 @@
         edge_types=PC_EDGE_TYPES,
         extensions=[PC_GRAPH_UI_EXTENSION],
         name="PC",
+        variables=[FloatVariable(name="threshold", value=PC_DEFAULT_THRESHOLD)],
     )
 )
 
 PCStable = graph_model_factory(
-    CausyAlgorithm(
+    Algorithm(
         pipeline_steps=[
             CalculatePearsonCorrelations(),
             ApplyActionsTogether(
                 pipeline_steps=[
-                    CorrelationCoefficientTest(threshold=PC_DEFAULT_THRESHOLD),
-                    PartialCorrelationTest(threshold=PC_DEFAULT_THRESHOLD),
+                    CorrelationCoefficientTest(
+                        threshold=VariableReference(name="threshold")
+                    ),
+                    PartialCorrelationTest(
+                        threshold=VariableReference(name="threshold"),
+                    ),
                     ExtendedPartialCorrelationTestMatrix(
-                        threshold=PC_DEFAULT_THRESHOLD
+                        threshold=VariableReference(name="threshold"),
                     ),
                 ]
             ),
@@ -99,27 +111,28 @@
         edge_types=PC_EDGE_TYPES,
         extensions=[PC_GRAPH_UI_EXTENSION],
         name="PCStable",
+        variables=[FloatVariable(name="threshold", value=PC_DEFAULT_THRESHOLD)],
     )
 )
 
 
 ParallelPC = graph_model_factory(
-    CausyAlgorithm(
+    Algorithm(
         pipeline_steps=[
             CalculatePearsonCorrelations(display_name="Calculate Pearson Correlations"),
             CorrelationCoefficientTest(
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 display_name="Correlation Coefficient Test",
             ),
             PartialCorrelationTest(
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 parallel=True,
                 chunk_size_parallel_processing=50000,
                 display_name="Partial Correlation Test",
             ),
             ExtendedPartialCorrelationTestMatrix(
                 # run first a sampled version of the test so we can minimize the number of tests in the full version
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 display_name="Sampled Extended Partial Correlation Test Matrix",
                 chunk_size_parallel_processing=5000,
                 parallel=True,
@@ -132,11 +145,11 @@
                         ),
                     ),
                     chunked=False,
-                    every_nth=200,
+                    every_nth=VariableReference(name="every_nth_sample"),
                 ),
             ),
             ExtendedPartialCorrelationTestMatrix(
-                threshold=PC_DEFAULT_THRESHOLD,
+                threshold=VariableReference(name="threshold"),
                 display_name="Extended Partial Correlation Test Matrix",
                 chunk_size_parallel_processing=20000,
                 parallel=True,
@@ -156,5 +169,9 @@
         edge_types=PC_EDGE_TYPES,
         extensions=[PC_GRAPH_UI_EXTENSION],
         name="ParallelPC",
+        variables=[
+            FloatVariable(name="threshold", value=PC_DEFAULT_THRESHOLD),
+            IntegerVariable(name="every_nth_sample", value=200),
+        ],
     )
 )

causy/independence_tests/common.py → causy/causal_discovery/constraint/independence_tests/common.py

@@ -10,13 +10,12 @@
     PipelineStepInterface,
     BaseGraphInterface,
     NodeInterface,
-    TestResult,
-    TestResultAction,
     AS_MANY_AS_FIELDS,
-    ComparisonSettings,
     GeneratorInterface,
     PipelineStepInterfaceType,
 )
+from causy.models import ComparisonSettings, TestResultAction, TestResult
+from causy.variables import IntegerParameter, BoolParameter
 
 logger = logging.getLogger(__name__)
 
@@ -27,8 +26,8 @@ class CorrelationCoefficientTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=2, max=2)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: List[str], graph: BaseGraphInterface
@@ -64,8 +63,8 @@ class PartialCorrelationTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=3, max=3)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: Tuple[str], graph: BaseGraphInterface
@@ -141,8 +140,8 @@ class ExtendedPartialCorrelationTestMatrix(
         comparison_settings=ComparisonSettings(min=4, max=AS_MANY_AS_FIELDS),
         shuffle_combinations=False,
     )
-    chunk_size_parallel_processing: int = 1000
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1000
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: List[str], graph: BaseGraphInterface
@@ -245,8 +244,8 @@ class ExtendedPartialCorrelationTestLinearRegression(
         comparison_settings=ComparisonSettings(min=4, max=AS_MANY_AS_FIELDS),
         shuffle_combinations=False,
     )
-    chunk_size_parallel_processing: int = 1000
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1000
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: List[str], graph: BaseGraphInterface

causy/orientation_rules/fci.py → causy/causal_discovery/constraint/orientation_rules/fci.py

@@ -2,14 +2,12 @@
 
 from causy.generators import AllCombinationsGenerator
 from causy.interfaces import (
-    TestResultAction,
     PipelineStepInterface,
-    ComparisonSettings,
     BaseGraphInterface,
-    TestResult,
     GeneratorInterface,
     PipelineStepInterfaceType,
 )
+from causy.models import ComparisonSettings, TestResultAction, TestResult
 
 
 class ColliderRuleFCI(

causy/orientation_rules/pc.py → causy/causal_discovery/constraint/orientation_rules/pc.py

@@ -4,13 +4,13 @@
 from causy.generators import AllCombinationsGenerator
 from causy.interfaces import (
     BaseGraphInterface,
-    TestResult,
-    TestResultAction,
     PipelineStepInterface,
-    ComparisonSettings,
     GeneratorInterface,
     PipelineStepInterfaceType,
 )
+from causy.models import ComparisonSettings, TestResultAction, TestResult
+from causy.variables import IntegerParameter, BoolParameter
+
 
 # theory for all orientation rules with pictures:
 # https://hpi.de/fileadmin/user_upload/fachgebiete/plattner/teaching/CausalInference/2019/Introduction_to_Constraint-Based_Causal_Structure_Learning.pdf
@@ -24,8 +24,8 @@ class ColliderTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=2, max=2)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: Tuple[str], graph: BaseGraphInterface
@@ -92,8 +92,8 @@ class NonColliderTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=2, max=2)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: Tuple[str], graph: BaseGraphInterface
@@ -158,8 +158,8 @@ class FurtherOrientTripleTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=2, max=2)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: Tuple[str], graph: BaseGraphInterface
@@ -216,8 +216,8 @@ class OrientQuadrupleTest(
     generator: Optional[GeneratorInterface] = AllCombinationsGenerator(
         comparison_settings=ComparisonSettings(min=2, max=2)
     )
-    chunk_size_parallel_processing: int = 1
-    parallel: bool = False
+    chunk_size_parallel_processing: IntegerParameter = 1
+    parallel: BoolParameter = False
 
     def process(
         self, nodes: Tuple[str], graph: BaseGraphInterface