added interface for statistic modules

medmodels/statistic_evaluations/statistical_analysis/descriptive_statistics.pyi

@@ -0,0 +1,18 @@
+from typing import Dict
+
+import polars as pl
+
+from medmodels.medrecord import MedRecord
+from medmodels.medrecord.querying import NodeOperation
+from medmodels.medrecord.schema import AttributeType
+
+def determine_attribute_type(attribute_values: pl.Series) -> AttributeType: ...
+def get_continuos_attribute_statistics(
+    medrecord: MedRecord, attribute_query: NodeOperation
+) -> Dict[str, float]: ...
+def get_temporal_attribute_statistics(
+    medrecord: MedRecord, attribute_query: NodeOperation
+) -> Dict[str, str]: ...
+def get_categorical_attribute_statistics(
+    medrecord: MedRecord, attribute_query: NodeOperation
+) -> Dict[str, str]: ...

medmodels/statistic_evaluations/statistical_analysis/inferential_statistics.pyi

@@ -0,0 +1,19 @@
+from typing import Dict, List, Union
+
+from numpy.typing import ArrayLike
+
+class TestHypothesis:
+    alpha: float
+    samples: List[ArrayLike]
+    sample_number: int
+    normal_distribution: bool
+    test: str
+    hypothesis: str
+    p_value: float
+
+    def test_for_normal_distribution(self) -> bool: ...
+    def decide_hypothesis_test(self) -> Dict[str, Union[str, bool, float]]: ...
+    def two_tailed_t_test(self) -> float: ...
+    def mann_whitney_u_test(self) -> float: ...
+    def ANOVA(self) -> float: ...
+    def chi_sqare_independece_test(self) -> float: ...

pyproject.toml

@@ -31,6 +31,7 @@ dependencies = [
     "pandas>=2.2.2",
     "polars[pandas]>=1.6.0",
     "scikit-learn>=1.5.0",
+    "scipy>=1.9.0",
 ]
 
 [project.optional-dependencies]