供应链因果决策图推理 — 从相关性到因果性，Palantir分析→行动的核心跨越

"""
供应链因果决策图推理系统
功能：DAG构建 / 后门调整 / 干预效果估计 / 反事实推理 / Palantir Action验证
输入：供应链数据 + 先验因果图结构
输出：因果效应估计 + 反事实分析 + 干预建议
"""
import numpy as np
import pandas as pd
from dataclasses import dataclass
from typing import Optional
import warnings
warnings.filterwarnings('ignore')


@dataclass
class CausalEffect:
    """因果效应估计结果——用于Palantir Action的理论依据"""
    treatment: str
    outcome: str
    ate: float                    # Average Treatment Effect
    ate_ci: tuple                 # 置信区间
    confounders_controlled: list  # 已控制的混杂变量
    identification_method: str    # 识别方法
    palantir_action_recommendation: str


class BackdoorAdjustmentEstimator:
    """
    后门调整法估计因果效应
    适用于：有混杂变量但DAG已知的场景
    """
    
    def __init__(self, confounders: list):
        self.confounders = confounders
        self._fitted = False
    
    def fit_estimate(self, data: pd.DataFrame,
                     treatment: str, outcome: str,
                     n_bootstrap: int = 200) -> CausalEffect:
        """
        后门调整估计因果效应
        E[Y | do(X=x)] = Σ_z E[Y|X=x, Z=z] * P(Z=z)
        """
        # 分层估计（对混杂变量进行条件化）
        results = []
        
        for _ in range(n_bootstrap):
            # Bootstrap重采样
            boot_data = data.sample(len(data), replace=True)
            
            ate = self._estimate_ate(boot_data, treatment, outcome)
            results.append(ate)
        
        ate_mean = np.mean(results)
        ate_ci = (np.percentile(results, 2.5), np.percentile(results, 97.5))
        
        self._fitted = True
        
        # 生成Palantir Action建议
        action_rec = self._generate_action_recommendation(