aigroup-econ-mcp 1.3.3__py3-none-any.whl → 2.0.1__py3-none-any.whl

tools/causal_inference_adapter.py

@@ -0,0 +1,658 @@
+"""
+因果推断方法适配器
+提供统一的接口调用econometrics/causal_inference中的各种因果识别方法
+"""
+
+from typing import List, Optional, Union, Dict, Any
+import json
+
+# 导入所有因果推断方法
+from econometrics.causal_inference.causal_identification_strategy.difference_in_differences import (
+    difference_in_differences, DIDResult
+)
+from econometrics.causal_inference.causal_identification_strategy.instrumental_variables import (
+    instrumental_variables_2sls, IVResult
+)
+from econometrics.causal_inference.causal_identification_strategy.propensity_score_matching import (
+    propensity_score_matching, PSMMatchResult
+)
+from econometrics.causal_inference.causal_identification_strategy.fixed_effects import (
+    fixed_effects_model, FixedEffectsResult
+)
+from econometrics.causal_inference.causal_identification_strategy.random_effects import (
+    random_effects_model, RandomEffectsResult
+)
+from econometrics.causal_inference.causal_identification_strategy.regression_discontinuity import (
+    regression_discontinuity, RDDResult
+)
+from econometrics.causal_inference.causal_identification_strategy.synthetic_control import (
+    synthetic_control_method, SyntheticControlResult
+)
+from econometrics.causal_inference.causal_identification_strategy.event_study import (
+    event_study, EventStudyResult
+)
+from econometrics.causal_inference.causal_identification_strategy.triple_difference import (
+    triple_difference, TripeDifferenceResult
+)
+from econometrics.causal_inference.causal_identification_strategy.mediation_analysis import (
+    mediation_analysis, MediationResult
+)
+from econometrics.causal_inference.causal_identification_strategy.moderation_analysis import (
+    moderation_analysis, ModerationResult
+)
+from econometrics.causal_inference.causal_identification_strategy.control_function import (
+    control_function_approach, ControlFunctionResult
+)
+from econometrics.causal_inference.causal_identification_strategy.first_difference import (
+    first_difference_model, FirstDifferenceResult
+)
+
+from .data_loader import DataLoader
+from .output_formatter import OutputFormatter
+
+
+def did_adapter(
+    treatment: Optional[List[int]] = None,
+    time_period: Optional[List[int]] = None,
+    outcome: Optional[List[float]] = None,
+    covariates: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    双重差分法 (DID) 适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            treatment = data.get("treatment", treatment)
+            time_period = data.get("time_period", time_period)
+            outcome = data.get("outcome", outcome)
+            covariates = data.get("covariates", covariates)
+        
+        # 调用核心方法
+        result: DIDResult = difference_in_differences(
+            treatment=treatment,
+            time_period=time_period,
+            outcome=outcome,
+            covariates=covariates
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"DID分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def iv_adapter(
+    y_data: Optional[List[float]] = None,
+    x_data: Optional[List[List[float]]] = None,
+    instruments: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    feature_names: Optional[List[str]] = None,
+    instrument_names: Optional[List[str]] = None,
+    constant: bool = True,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    工具变量法 (IV/2SLS) 适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            y_data = data.get("y_data", y_data)
+            x_data = data.get("x_data", x_data)
+            instruments = data.get("instruments", instruments)
+        
+        # 调用核心方法
+        result: IVResult = instrumental_variables_2sls(
+            y=y_data,
+            x=x_data,
+            instruments=instruments,
+            feature_names=feature_names,
+            instrument_names=instrument_names,
+            constant=constant
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"IV/2SLS分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def psm_adapter(
+    treatment: Optional[List[int]] = None,
+    outcome: Optional[List[float]] = None,
+    covariates: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    matching_method: str = "nearest",
+    k_neighbors: int = 1,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    倾向得分匹配 (PSM) 适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            treatment = data.get("treatment", treatment)
+            outcome = data.get("outcome", outcome)
+            covariates = data.get("covariates", covariates)
+        
+        # 调用核心方法
+        result: PSMMatchResult = propensity_score_matching(
+            treatment=treatment,
+            outcome=outcome,
+            covariates=covariates,
+            matching_method=matching_method,
+            k_neighbors=k_neighbors
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"PSM分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def fixed_effects_adapter(
+    y_data: Optional[List[float]] = None,
+    x_data: Optional[List[List[float]]] = None,
+    entity_ids: Optional[List[str]] = None,
+    time_periods: Optional[List[str]] = None,
+    file_path: Optional[str] = None,
+    constant: bool = True,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    固定效应模型适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            y_data = data.get("y_data", y_data)
+            x_data = data.get("x_data", x_data)
+            entity_ids = data.get("entity_ids", entity_ids)
+            time_periods = data.get("time_periods", time_periods)
+        
+        # 调用核心方法
+        result: FixedEffectsResult = fixed_effects_model(
+            y=y_data,
+            x=x_data,
+            entity_ids=entity_ids,
+            time_periods=time_periods,
+            constant=constant
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"固定效应模型分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def random_effects_adapter(
+    y_data: Optional[List[float]] = None,
+    x_data: Optional[List[List[float]]] = None,
+    entity_ids: Optional[List[str]] = None,
+    time_periods: Optional[List[str]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    随机效应模型适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            y_data = data.get("y_data", y_data)
+            x_data = data.get("x_data", x_data)
+            entity_ids = data.get("entity_ids", entity_ids)
+            time_periods = data.get("time_periods", time_periods)
+        
+        # 调用核心方法
+        result: RandomEffectsResult = random_effects_model(
+            y=y_data,
+            x=x_data,
+            entity_ids=entity_ids,
+            time_periods=time_periods
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"随机效应模型分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def rdd_adapter(
+    running_variable: Optional[List[float]] = None,
+    outcome: Optional[List[float]] = None,
+    cutoff: float = 0.0,
+    file_path: Optional[str] = None,
+    bandwidth: Optional[float] = None,
+    polynomial_order: int = 1,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    回归断点设计 (RDD) 适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            running_variable = data.get("running_variable", running_variable)
+            outcome = data.get("outcome", outcome)
+            cutoff = data.get("cutoff", cutoff)
+        
+        # 调用核心方法
+        result: RDDResult = regression_discontinuity(
+            running_variable=running_variable,
+            outcome=outcome,
+            cutoff=cutoff,
+            bandwidth=bandwidth,
+            polynomial_order=polynomial_order
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"RDD分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def synthetic_control_adapter(
+    outcome: Optional[List[float]] = None,
+    treatment_period: int = 0,
+    treated_unit: str = "unit_1",
+    donor_units: Optional[List[str]] = None,
+    time_periods: Optional[List[str]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    合成控制法适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            outcome = data.get("outcome", outcome)
+            treatment_period = data.get("treatment_period", treatment_period)
+            treated_unit = data.get("treated_unit", treated_unit)
+            donor_units = data.get("donor_units", donor_units)
+            time_periods = data.get("time_periods", time_periods)
+        
+        # 调用核心方法
+        result: SyntheticControlResult = synthetic_control_method(
+            outcome=outcome,
+            treatment_period=treatment_period,
+            treated_unit=treated_unit,
+            donor_units=donor_units,
+            time_periods=time_periods
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"合成控制法分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def event_study_adapter(
+    outcome: Optional[List[float]] = None,
+    treatment: Optional[List[int]] = None,
+    entity_ids: Optional[List[str]] = None,
+    time_periods: Optional[List[str]] = None,
+    event_time: Optional[List[int]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    事件研究法适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            outcome = data.get("outcome", outcome)
+            treatment = data.get("treatment", treatment)
+            entity_ids = data.get("entity_ids", entity_ids)
+            time_periods = data.get("time_periods", time_periods)
+            event_time = data.get("event_time", event_time)
+        
+        # 调用核心方法
+        result: EventStudyResult = event_study(
+            outcome=outcome,
+            treatment=treatment,
+            entity_ids=entity_ids,
+            time_periods=time_periods,
+            event_time=event_time
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"事件研究法分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def triple_difference_adapter(
+    outcome: Optional[List[float]] = None,
+    treatment_group: Optional[List[int]] = None,
+    time_period: Optional[List[int]] = None,
+    cohort_group: Optional[List[int]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    三重差分法 (DDD) 适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            outcome = data.get("outcome", outcome)
+            treatment_group = data.get("treatment_group", treatment_group)
+            time_period = data.get("time_period", time_period)
+            cohort_group = data.get("cohort_group", cohort_group)
+        
+        # 调用核心方法
+        result: TripeDifferenceResult = triple_difference(
+            outcome=outcome,
+            treatment_group=treatment_group,
+            time_period=time_period,
+            cohort_group=cohort_group
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"DDD分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def mediation_adapter(
+    outcome: Optional[List[float]] = None,
+    treatment: Optional[List[float]] = None,
+    mediator: Optional[List[float]] = None,
+    covariates: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    中介效应分析适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            outcome = data.get("outcome", outcome)
+            treatment = data.get("treatment", treatment)
+            mediator = data.get("mediator", mediator)
+            covariates = data.get("covariates", covariates)
+        
+        # 调用核心方法
+        result: MediationResult = mediation_analysis(
+            outcome=outcome,
+            treatment=treatment,
+            mediator=mediator,
+            covariates=covariates
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"中介效应分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def moderation_adapter(
+    outcome: Optional[List[float]] = None,
+    predictor: Optional[List[float]] = None,
+    moderator: Optional[List[float]] = None,
+    covariates: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    调节效应分析适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            outcome = data.get("outcome", outcome)
+            predictor = data.get("predictor", predictor)
+            moderator = data.get("moderator", moderator)
+            covariates = data.get("covariates", covariates)
+        
+        # 调用核心方法
+        result: ModerationResult = moderation_analysis(
+            outcome=outcome,
+            predictor=predictor,
+            moderator=moderator,
+            covariates=covariates
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"调节效应分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def control_function_adapter(
+    y_data: Optional[List[float]] = None,
+    x_data: Optional[List[float]] = None,
+    z_data: Optional[List[List[float]]] = None,
+    file_path: Optional[str] = None,
+    constant: bool = True,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    控制函数法适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            y_data = data.get("y_data", y_data)
+            x_data = data.get("x_data", x_data)
+            z_data = data.get("z_data", z_data)
+        
+        # 调用核心方法
+        result: ControlFunctionResult = control_function_approach(
+            y=y_data,
+            x=x_data,
+            z=z_data,
+            constant=constant
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"控制函数法分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)
+
+
+def first_difference_adapter(
+    y_data: Optional[List[float]] = None,
+    x_data: Optional[List[float]] = None,
+    entity_ids: Optional[List[str]] = None,
+    file_path: Optional[str] = None,
+    output_format: str = "json",
+    save_path: Optional[str] = None
+) -> str:
+    """
+    一阶差分模型适配器
+    """
+    try:
+        # 从文件加载数据
+        if file_path:
+            data = DataLoader.load_from_file(file_path)
+            y_data = data.get("y_data", y_data)
+            x_data = data.get("x_data", x_data)
+            entity_ids = data.get("entity_ids", entity_ids)
+        
+        # 调用核心方法
+        result: FirstDifferenceResult = first_difference_model(
+            y=y_data,
+            x=x_data,
+            entity_ids=entity_ids
+        )
+        
+        # 格式化输出
+        if output_format == "json":
+            output = result.model_dump_json(indent=2)
+        else:
+            output = str(result.model_dump())
+        
+        # 保存结果
+        if save_path:
+            OutputFormatter.save_to_file(output, save_path)
+        
+        return output
+        
+    except Exception as e:
+        error_msg = f"一阶差分模型分析错误: {str(e)}"
+        return json.dumps({"error": error_msg}, indent=2)