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

tools/README.md

@@ -0,0 +1,88 @@
+# Tools Directory
+
+## 当前架构 (v2.0 - Adapter Pattern)
+
+### 活跃文件
+
+1. **econometrics_adapter.py** - 核心适配器
+   - 将 econometrics/ 核心算法适配为 MCP 工具
+   - 支持文件输入和多种输出格式
+   - 减少 84% 代码重复
+
+2. **data_loader.py** - 数据加载组件
+   - 支持 txt/json/csv/excel 格式
+   - DataLoader: OLS/GMM 数据
+   - MLEDataLoader: MLE 数据
+
+3. **output_formatter.py** - 输出格式化组件
+   - MarkdownFormatter: Markdown 格式
+   - TextFormatter: 纯文本格式
+   - 结果保存功能
+
+### 架构优势
+
+```
+MCP Server (fastmcp_server.py)
+    ↓
+Adapter Layer (econometrics_adapter.py) - 90 lines
+    ↓
+Core Algorithms (econometrics/basic_parametric_estimation/)
+    ├── ols/ols_model.py - OLS 核心算法
+    ├── mle/mle_model.py - MLE 核心算法
+    └── gmm/gmm_model.py - GMM 核心算法 (已修复 j_p_value bug)
+```
+
+**优点**：
+- ✅ 代码复用：单一真相源
+- ✅ DRY 原则：不重复自己
+- ✅ 易于维护：Bug 只需修复一次
+- ✅ 清晰分层：职责明确
+- ✅ 易于扩展：新算法只需写适配器
+
+### 备份文件
+
+`bak_old_implementation/` - 旧的独立实现（已废弃）
+- ols_tool.py (155 lines)
+- mle_tool.py (219 lines) 
+- gmm_tool.py (190 lines)
+
+这些文件已被 econometrics_adapter.py (164 lines) 替代
+节省代码：474 lines (84%)
+
+## 使用示例
+
+```python
+from tools.econometrics_adapter import (
+    ols_adapter,
+    mle_adapter,
+    gmm_adapter
+)
+
+# 直接数据输入
+result = ols_adapter(
+    y_data=[1,2,3,4,5],
+    x_data=[[1],[2],[3],[4],[5]],
+    output_format="json"
+)
+
+# 文件输入
+result = mle_adapter(
+    file_path="data/sample.csv",
+    distribution="normal",
+    output_format="markdown",
+    save_path="results/mle.md"
+)
+```
+
+## 迁移记录
+
+- 2025-11-04: 切换到适配器模式
+- 修复核心 GMM bug (j_p_value)
+- 移动旧实现到 bak_old_implementation/
+- 减少 84% 重复代码
+
+## 相关文档
+
+- [架构分析](../ARCHITECTURE_ANALYSIS.md)
+- [服务器代码](../fastmcp_server.py)
+- [核心算法](../econometrics/basic_parametric_estimation/)

tools/__init__.py

@@ -0,0 +1,119 @@
+"""
+工具模块初始化文件
+"""
+
+from .data_loader import DataLoader
+from .output_formatter import OutputFormatter
+from .econometrics_adapter import EconometricsAdapter
+
+# 时间序列和面板数据工具
+from .time_series_panel_data_adapter import TimeSeriesPanelDataAdapter
+from .time_series_panel_data_tools import (
+    arima_model,
+    exponential_smoothing_model,
+    garch_model,
+    unit_root_tests,
+    var_svar_model,
+    cointegration_analysis,
+    dynamic_panel_model
+)
+
+# 因果推断工具适配器
+from .causal_inference_adapter import (
+    did_adapter,
+    iv_adapter,
+    psm_adapter,
+    fixed_effects_adapter,
+    random_effects_adapter,
+    rdd_adapter,
+    synthetic_control_adapter,
+    event_study_adapter,
+    triple_difference_adapter,
+    mediation_adapter,
+    moderation_adapter,
+    control_function_adapter,
+    first_difference_adapter
+)
+
+# 机器学习工具适配器
+from .machine_learning_adapter import (
+    random_forest_adapter,
+    gradient_boosting_adapter,
+    svm_adapter,
+    neural_network_adapter,
+    kmeans_clustering_adapter,
+    hierarchical_clustering_adapter,
+    double_ml_adapter,
+    causal_forest_adapter
+)
+
+# 微观计量模型工具适配器
+from .microecon_adapter import (
+    logit_adapter,
+    probit_adapter,
+    multinomial_logit_adapter,
+    poisson_adapter,
+    negative_binomial_adapter,
+    tobit_adapter,
+    heckman_adapter
+)
+
+# 保持向后兼容性
+ols_adapter = EconometricsAdapter.ols_regression
+mle_adapter = EconometricsAdapter.mle_estimation
+gmm_adapter = EconometricsAdapter.gmm_estimation
+
+__all__ = [
+    "DataLoader",
+    "OutputFormatter",
+    "EconometricsAdapter",
+    "TimeSeriesPanelDataAdapter",
+    
+    # 基础工具
+    "ols_adapter",
+    "mle_adapter",
+    "gmm_adapter",
+    
+    # 时间序列和面板数据工具
+    "arima_model",
+    "exponential_smoothing_model",
+    "garch_model",
+    "unit_root_tests",
+    "var_svar_model",
+    "cointegration_analysis",
+    "dynamic_panel_model",
+    
+    # 因果推断工具
+    "did_adapter",
+    "iv_adapter",
+    "psm_adapter",
+    "fixed_effects_adapter",
+    "random_effects_adapter",
+    "rdd_adapter",
+    "synthetic_control_adapter",
+    "event_study_adapter",
+    "triple_difference_adapter",
+    "mediation_adapter",
+    "moderation_adapter",
+    "control_function_adapter",
+    "first_difference_adapter",
+    
+    # 机器学习工具
+    "random_forest_adapter",
+    "gradient_boosting_adapter",
+    "svm_adapter",
+    "neural_network_adapter",
+    "kmeans_clustering_adapter",
+    "hierarchical_clustering_adapter",
+    "double_ml_adapter",
+    "causal_forest_adapter",
+    
+    # 微观计量模型工具
+    "logit_adapter",
+    "probit_adapter",
+    "multinomial_logit_adapter",
+    "poisson_adapter",
+    "negative_binomial_adapter",
+    "tobit_adapter",
+    "heckman_adapter"
+]