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

econometrics/tests/causal_inference_tests/test_instrumental_variables.py

@@ -0,0 +1,44 @@
+"""
+工具变量法测试
+"""
+
+import numpy as np
+import unittest
+from econometrics.causal_inference.causal_identification_strategy import instrumental_variables_2sls
+
+
+class TestInstrumentalVariables(unittest.TestCase):
+    
+    def test_instrumental_variables_2sls(self):
+        """测试工具变量法"""
+        # 生成模拟数据
+        np.random.seed(42)
+        n = 100
+        
+        # 工具变量
+        z = np.random.normal(0, 1, n)
+        
+        # 内生变量（与误差项相关）
+        e1 = np.random.normal(0, 1, n)
+        x = 1 + 0.5 * z + e1
+        
+        # 结果变量
+        e2 = np.random.normal(0, 1, n)
+        y = 2 + 1.5 * x + e2 + 0.3 * e1  # 包含内生性
+        
+        # 执行工具变量回归
+        result = instrumental_variables_2sls(
+            y=y.tolist(),
+            x=x.reshape(-1, 1).tolist(),
+            instruments=z.reshape(-1, 1).tolist()
+        )
+        
+        # 检查结果
+        self.assertIsNotNone(result.estimate)
+        self.assertIsNotNone(result.std_error)
+        self.assertIsNotNone(result.p_value)
+        self.assertGreater(result.n_observations, 0)
+
+
+if __name__ == "__main__":
+    unittest.main()

econometrics/tests/model_specification_diagnostics_tests/__init__.py

@@ -0,0 +1,3 @@
+"""
+模型设定、诊断与稳健推断测试模块初始化文件
+"""

econometrics/tests/model_specification_diagnostics_tests/test_diagnostic_tests.py

@@ -0,0 +1,86 @@
+"""
+诊断检验模型测试脚本
+"""
+
+import sys
+import os
+import numpy as np
+
+# 添加项目根目录到路径
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+
+# 尝试导入诊断检验模型
+try:
+    from econometrics.model_specification_diagnostics_robust_inference.diagnostic_tests.diagnostic_tests_model import diagnostic_tests, DiagnosticTestsResult
+    DIAGNOSTIC_TESTS_AVAILABLE = True
+except ImportError:
+    DIAGNOSTIC_TESTS_AVAILABLE = False
+    print("警告: 未找到诊断检验模型，相关测试将被跳过")
+
+
+def test_diagnostic_tests_basic():
+    """测试基本诊断检验功能"""
+    if not DIAGNOSTIC_TESTS_AVAILABLE:
+        print("跳过诊断检验测试（模块不可用）")
+        return
+        
+    print("测试基本诊断检验功能...")
+    
+    # 生成测试数据
+    np.random.seed(42)
+    n = 100
+    x1 = np.random.randn(n)
+    x2 = np.random.randn(n)
+    # 真实模型: y = 2 + 3*x1 + 2*x2 + noise
+    y = 2 + 3*x1 + 2*x2 + np.random.randn(n) * 0.5
+    
+    # 准备数据
+    x_data = [[x1[i], x2[i]] for i in range(n)]
+    y_data = y.tolist()
+    
+    # 执行诊断检验
+    result = diagnostic_tests(y_data, x_data, feature_names=['x1', 'x2'])
+    
+    # 验证结果类型
+    assert isinstance(result, DiagnosticTestsResult), "结果应为DiagnosticTestsResult类型"
+    
+    # 验证统计量合理性
+    assert result.n_obs == n, f"观测数量应为{n}"
+    
+    print("  Jarque-Bera检验统计量:", result.jarque_bera_stat)
+    print("  Breusch-Pagan检验统计量:", result.breusch_pagan_stat)
+    print("  基本诊断检验功能测试通过")
+
+
+def test_diagnostic_tests_serial_correlation():
+    """测试序列相关诊断检验"""
+    if not DIAGNOSTIC_TESTS_AVAILABLE:
+        print("跳过序列相关诊断检验测试（模块不可用）")
+        return
+        
+    print("测试序列相关诊断检验...")
+    
+    # 生成带序列相关的测试数据
+    np.random.seed(42)
+    n = 100
+    # 生成自相关误差项
+    errors = [0]
+    for i in range(1, n):
+        errors.append(0.5 * errors[i-1] + np.random.randn() * 0.3)
+    
+    x = np.random.randn(n)
+    # 真实模型: y = 2 + 3*x + errors (带序列相关)
+    y = 2 + 3*x + np.array(errors)
+    
+    # 执行诊断检验
+    result = diagnostic_tests(y.tolist(), x.tolist(), feature_names=['x1'])
+    
+    print("  Durbin-Watson统计量:", result.durbin_watson_stat)
+    print("  序列相关诊断检验测试通过")
+
+
+if __name__ == "__main__":
+    print("开始测试诊断检验模型...")
+    test_diagnostic_tests_basic()
+    test_diagnostic_tests_serial_correlation()
+    print("诊断检验测试完成!")

econometrics/tests/model_specification_diagnostics_tests/test_robust_errors.py

@@ -0,0 +1,89 @@
+"""
+稳健标准误模型测试脚本
+"""
+
+import sys
+import os
+import numpy as np
+
+# 添加项目根目录到路径
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+
+# 尝试导入稳健标准误模型
+try:
+    from econometrics.model_specification_diagnostics_robust_inference.robust_errors.robust_errors_model import robust_errors_regression, RobustErrorsResult
+    ROBUST_ERRORS_AVAILABLE = True
+except ImportError:
+    ROBUST_ERRORS_AVAILABLE = False
+    print("警告: 未找到稳健标准误模型，相关测试将被跳过")
+
+
+def test_robust_errors_basic():
+    """测试基本稳健标准误功能"""
+    if not ROBUST_ERRORS_AVAILABLE:
+        print("跳过稳健标准误测试（模块不可用）")
+        return
+        
+    print("测试基本稳健标准误功能...")
+    
+    # 生成测试数据
+    np.random.seed(42)
+    n = 100
+    x1 = np.random.randn(n)
+    x2 = np.random.randn(n)
+    # 真实模型: y = 2 + 3*x1 + 2*x2 + noise (异方差)
+    noise = np.random.randn(n) * (0.5 + 0.3 * np.abs(x1))  # 异方差噪声
+    y = 2 + 3*x1 + 2*x2 + noise
+    
+    # 准备数据
+    x_data = [[x1[i], x2[i]] for i in range(n)]
+    y_data = y.tolist()
+    
+    # 执行稳健标准误回归
+    result = robust_errors_regression(y_data, x_data, feature_names=['x1', 'x2'])
+    
+    # 验证结果类型
+    assert isinstance(result, RobustErrorsResult), "结果应为RobustErrorsResult类型"
+    
+    # 验证系数数量
+    assert len(result.coefficients) == 3, "应该有3个系数（包括常数项）"
+    
+    # 验证统计量合理性
+    assert result.n_obs == n, f"观测数量应为{n}"
+    
+    print("  系数:", result.coefficients)
+    print("  稳健标准误:", result.robust_std_errors)
+    print("  基本稳健标准误功能测试通过")
+
+
+def test_robust_errors_no_constant():
+    """测试不包含常数项的稳健标准误"""
+    if not ROBUST_ERRORS_AVAILABLE:
+        print("跳过稳健标准误测试（模块不可用）")
+        return
+        
+    print("测试不包含常数项的稳健标准误...")
+    
+    # 生成测试数据
+    np.random.seed(42)
+    n = 50
+    x = np.random.randn(n)
+    # 真实模型: y = 2*x + noise (异方差)
+    noise = np.random.randn(n) * (0.3 + 0.2 * np.abs(x))  # 异方差噪声
+    y = 2*x + noise
+    
+    # 执行稳健标准误回归（不包含常数项）
+    result = robust_errors_regression(y.tolist(), x.tolist(), constant=False, feature_names=['x1'])
+    
+    # 验证系数数量
+    assert len(result.coefficients) == 1, "应该有1个系数（不包括常数项）"
+    
+    print("  系数:", result.coefficients)
+    print("  不包含常数项的稳健标准误测试通过")
+
+
+if __name__ == "__main__":
+    print("开始测试稳健标准误模型...")
+    test_robust_errors_basic()
+    test_robust_errors_no_constant()
+    print("稳健标准误测试完成!")

econometrics/tests/specific_data_modeling_tests/__init__.py

@@ -0,0 +1,3 @@
+"""
+特定数据建模测试模块初始化文件
+"""

econometrics/tests/specific_data_modeling_tests/test_arima.py

@@ -0,0 +1,98 @@
+"""
+ARIMA模型测试脚本
+"""
+
+import sys
+import os
+import numpy as np
+
+# 添加项目根目录到路径
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+
+from econometrics.specific_data_modeling.time_series_panel_data.arima_model import arima_model, ARIMAResult
+
+
+def test_arima_basic():
+    """测试基本ARIMA功能"""
+    print("测试基本ARIMA功能...")
+    
+    # 生成AR(1)过程数据
+    np.random.seed(42)
+    n = 100
+    ar_param = 0.6
+    data = [0]  # 初始值
+    
+    # 生成AR(1)过程: y[t] = 0.6 * y[t-1] + noise
+    for i in range(1, n):
+        data.append(ar_param * data[i-1] + np.random.randn() * 0.5)
+    
+    # 执行ARIMA模型 (1,0,0) 即 AR(1)
+    result = arima_model(data, order=(1, 0, 0), forecast_steps=5)
+    
+    # 验证结果类型
+    assert isinstance(result, ARIMAResult), "结果应为ARIMAResult类型"
+    
+    # 验证模型类型
+    assert "ARIMA(1,0,0)" in result.model_type, "模型类型应为ARIMA(1,0,0)"
+    
+    # 验证系数数量
+    assert len(result.coefficients) >= 1, "应该至少有1个系数"
+    
+    # 验证统计量合理性
+    assert result.n_obs == n, f"观测数量应为{n}"
+    assert len(result.forecast) == 5, "预测步数应为5"
+    
+    print("  模型类型:", result.model_type)
+    print("  系数:", result.coefficients)
+    print("  AIC:", result.aic)
+    print("  预测值:", result.forecast[:3], "...")
+    print("  基本ARIMA功能测试通过")
+
+
+def test_arima_integrated():
+    """测试带积分的ARIMA模型"""
+    print("测试带积分的ARIMA模型...")
+    
+    # 生成带趋势的随机游走数据
+    np.random.seed(42)
+    n = 80
+    data = [0]  # 初始值
+    
+    # 生成随机游走过程: y[t] = y[t-1] + noise
+    for i in range(1, n):
+        data.append(data[i-1] + np.random.randn() * 0.3)
+    
+    # 添加线性趋势
+    data = [data[i] + 0.1 * i for i in range(len(data))]
+    
+    # 执行ARIMA模型 (0,1,0) 即随机游走
+    result = arima_model(data, order=(0, 1, 0), forecast_steps=3)
+    
+    # 验证模型类型
+    assert "ARIMA(0,1,0)" in result.model_type, "模型类型应为ARIMA(0,1,0)"
+    
+    print("  模型类型:", result.model_type)
+    print("  预测值:", result.forecast)
+    print("  带积分的ARIMA模型测试通过")
+
+
+def test_arima_errors():
+    """测试ARIMA错误处理"""
+    print("测试ARIMA错误处理...")
+    
+    # 测试空数据
+    try:
+        arima_model([])
+        assert False, "应该抛出异常"
+    except Exception:
+        print("  空数据错误处理正确")
+    
+    print("  ARIMA错误处理测试通过")
+
+
+if __name__ == "__main__":
+    print("开始测试ARIMA模型...")
+    test_arima_basic()
+    test_arima_integrated()
+    test_arima_errors()
+    print("所有ARIMA测试通过!")

econometrics/tests/specific_data_modeling_tests/test_dynamic_panel.py

@@ -0,0 +1,198 @@
+"""
+动态面板模型测试脚本
+"""
+
+import sys
+import os
+import numpy as np
+
+# 添加项目根目录到路径
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+
+# 尝试导入动态面板模型
+try:
+    from econometrics.specific_data_modeling.time_series_panel_data.dynamic_panel_models import diff_gmm_model, sys_gmm_model, DynamicPanelResult
+    DYNAMIC_PANEL_AVAILABLE = True
+except ImportError:
+    DYNAMIC_PANEL_AVAILABLE = False
+    print("警告: 未找到动态面板模型，相关测试将被跳过")
+
+
+def test_diff_gmm_model():
+    """测试差分GMM模型"""
+    if not DYNAMIC_PANEL_AVAILABLE:
+        print("跳过差分GMM测试（模块不可用）")
+        return
+        
+    print("测试差分GMM模型...")
+    
+    # 生成面板数据
+    np.random.seed(42)
+    n_individuals = 20  # 个体数
+    n_time_periods = 15  # 时间期数
+    
+    # 生成个体和时间标识符
+    entity_ids = []
+    time_periods = []
+    
+    # 生成数据
+    y_data = []
+    x1_data = []
+    x2_data = []
+    
+    # 为每个个体生成数据
+    for i in range(n_individuals):
+        # 个体固定效应
+        entity_effect = np.random.randn() * 0.5
+        
+        # 初始值
+        y_prev = np.random.randn()
+        
+        for t in range(n_time_periods):
+            entity_ids.append(i)
+            time_periods.append(t)
+            
+            # 生成解释变量
+            x1 = np.random.randn()
+            x2 = np.random.randn()
+            
+            # 动态面板模型: y[i,t] = 0.5 * y[i,t-1] + 1.2 * x1 + 0.8 * x2 + entity_effect + noise
+            # 对于t=0，使用上一个个体的最后一个值或随机值
+            if t == 0:
+                y = 0.5 * y_prev + 1.2 * x1 + 0.8 * x2 + entity_effect + np.random.randn() * 0.5
+            else:
+                # 使用滞后因变量
+                y = 0.5 * y_data[-1] + 1.2 * x1 + 0.8 * x2 + entity_effect + np.random.randn() * 0.5
+            
+            y_data.append(y)
+            x1_data.append(x1)
+            x2_data.append(x2)
+            
+            y_prev = y
+    
+    # 准备自变量数据
+    x_data = [x1_data, x2_data]
+    
+    # 执行差分GMM模型
+    try:
+        result = diff_gmm_model(y_data, x_data, entity_ids, time_periods, lags=1)
+        
+        # 验证结果类型
+        assert isinstance(result, DynamicPanelResult), "结果应为DynamicPanelResult类型"
+        
+        # 验证模型类型
+        assert "Difference GMM" in result.model_type, "模型类型应为差分GMM"
+        
+        # 验证统计量合理性
+        assert result.n_obs == len(y_data), f"观测数量应为{len(y_data)}"
+        assert result.n_individuals == n_individuals, f"个体数量应为{n_individuals}"
+        assert result.n_time_periods == n_time_periods, f"时间期数应为{n_time_periods}"
+        
+        print("  模型类型:", result.model_type)
+        print("  系数数量:", len(result.coefficients))
+        print("  系数:", result.coefficients)
+        print("  J统计量:", result.j_statistic)
+        print("  差分GMM模型测试通过")
+        
+    except Exception as e:
+        print(f"  差分GMM模型测试跳过（可能需要更多数据或特定配置）: {e}")
+
+
+def test_sys_gmm_model():
+    """测试系统GMM模型"""
+    if not DYNAMIC_PANEL_AVAILABLE:
+        print("跳过系统GMM测试（模块不可用）")
+        return
+        
+    print("测试系统GMM模型...")
+    
+    # 生成面板数据
+    np.random.seed(42)
+    n_individuals = 15  # 个体数
+    n_time_periods = 12  # 时间期数
+    
+    # 生成个体和时间标识符
+    entity_ids = []
+    time_periods = []
+    
+    # 生成数据
+    y_data = []
+    x_data_list = [[] for _ in range(2)]  # 2个自变量
+    
+    # 为每个个体生成数据
+    for i in range(n_individuals):
+        # 个体固定效应
+        entity_effect = np.random.randn() * 0.5
+        
+        # 初始值
+        y_prev = np.random.randn()
+        
+        for t in range(n_time_periods):
+            entity_ids.append(i)
+            time_periods.append(t)
+            
+            # 生成解释变量
+            x1 = np.random.randn()
+            x2 = np.random.randn()
+            
+            # 动态面板模型: y[i,t] = 0.4 * y[i,t-1] + 1.0 * x1 + 0.7 * x2 + entity_effect + noise
+            if t == 0:
+                y = 0.4 * y_prev + 1.0 * x1 + 0.7 * x2 + entity_effect + np.random.randn() * 0.5
+            else:
+                # 使用滞后因变量
+                y = 0.4 * y_data[-1] + 1.0 * x1 + 0.7 * x2 + entity_effect + np.random.randn() * 0.5
+            
+            y_data.append(y)
+            x_data_list[0].append(x1)
+            x_data_list[1].append(x2)
+            
+            y_prev = y
+    
+    # 执行系统GMM模型
+    try:
+        result = sys_gmm_model(y_data, x_data_list, entity_ids, time_periods, lags=1)
+        
+        # 验证结果类型
+        assert isinstance(result, DynamicPanelResult), "结果应为DynamicPanelResult类型"
+        
+        # 验证模型类型
+        assert "System GMM" in result.model_type, "模型类型应为系统GMM"
+        
+        print("  模型类型:", result.model_type)
+        print("  系数:", result.coefficients)
+        print("  系统GMM模型测试通过")
+        
+    except Exception as e:
+        print(f"  系统GMM模型测试跳过（可能需要更多数据或特定配置）: {e}")
+
+
+def test_dynamic_panel_errors():
+    """测试动态面板模型错误处理"""
+    if not DYNAMIC_PANEL_AVAILABLE:
+        print("跳过动态面板错误处理测试（模块不可用）")
+        return
+        
+    print("测试动态面板模型错误处理...")
+    
+    # 测试空数据
+    try:
+        diff_gmm_model([], [[], []], [], [])
+        assert False, "应该抛出异常"
+    except Exception:
+        print("  差分GMM空数据错误处理正确")
+    
+    try:
+        sys_gmm_model([], [[], []], [], [])
+        assert False, "应该抛出异常"
+    except Exception:
+        print("  系统GMM空数据错误处理正确")
+    
+    print("  动态面板模型错误处理测试通过")
+
+
+if __name__ == "__main__":
+    print("开始测试动态面板模型...")
+    test_diff_gmm_model()
+    test_sys_gmm_model()
+    test_dynamic_panel_errors()
+    print("动态面板模型测试完成!")

econometrics/tests/specific_data_modeling_tests/test_exponential_smoothing.py

@@ -0,0 +1,105 @@
+"""
+指数平滑模型测试脚本
+"""
+
+import sys
+import os
+import numpy as np
+
+# 添加项目根目录到路径
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', '..'))
+
+from econometrics.specific_data_modeling.time_series_panel_data.exponential_smoothing import exponential_smoothing_model, ExponentialSmoothingResult
+
+
+def test_exponential_smoothing_basic():
+    """测试基本指数平滑功能"""
+    print("测试基本指数平滑功能...")
+    
+    # 生成带趋势的时间序列数据
+    np.random.seed(42)
+    n = 100
+    trend = 0.3
+    data = []
+    
+    # 生成带线性趋势和噪声的数据
+    for i in range(n):
+        value = 10 + trend * i + np.random.randn() * 2
+        data.append(value)
+    
+    # 执行指数平滑模型（带趋势）
+    result = exponential_smoothing_model(data, trend=True, seasonal=False, forecast_steps=5)
+    
+    # 验证结果类型
+    assert isinstance(result, ExponentialSmoothingResult), "结果应为ExponentialSmoothingResult类型"
+    
+    # 验证模型类型
+    assert "Exponential Smoothing with Trend" in result.model_type, "模型类型应包含趋势成分"
+    
+    # 验证统计量合理性
+    assert result.n_obs == n, f"观测数量应为{n}"
+    assert len(result.coefficients) >= 1, "应至少有1个系数"
+    assert len(result.forecast) == 5, "预测步数应为5"
+    
+    print("  模型类型:", result.model_type)
+    print("  系数数量:", len(result.coefficients))
+    print("  系数:", result.coefficients[:3], "...")
+    print("  平滑水平参数:", result.smoothing_level)
+    print("  平滑趋势参数:", result.smoothing_trend)
+    print("  MSE:", result.mse)
+    print("  预测值:", result.forecast[:3], "...")
+    print("  基本指数平滑功能测试通过")
+
+
+def test_exponential_smoothing_simple():
+    """测试简单指数平滑（无趋势）"""
+    print("测试简单指数平滑（无趋势）...")
+    
+    # 生成平稳时间序列数据
+    np.random.seed(42)
+    n = 80
+    mean_value = 5
+    data = []
+    
+    # 生成平稳数据
+    for i in range(n):
+        value = mean_value + np.random.randn() * 1.5
+        data.append(value)
+    
+    # 执行简单指数平滑模型（无趋势）
+    result = exponential_smoothing_model(data, trend=False, seasonal=False, forecast_steps=3)
+    
+    # 验证模型类型
+    assert "Exponential Smoothing" == result.model_type, "模型类型应为简单指数平滑"
+    
+    print("  模型类型:", result.model_type)
+    print("  简单指数平滑测试通过")
+
+
+def test_exponential_smoothing_errors():
+    """测试指数平滑错误处理"""
+    print("测试指数平滑错误处理...")
+    
+    # 测试空数据
+    try:
+        exponential_smoothing_model([])
+        assert False, "应该抛出异常"
+    except Exception:
+        print("  空数据错误处理正确")
+    
+    # 测试数据过少
+    try:
+        exponential_smoothing_model([1])
+        assert False, "应该抛出异常"
+    except Exception:
+        print("  数据过少错误处理正确")
+    
+    print("  指数平滑错误处理测试通过")
+
+
+if __name__ == "__main__":
+    print("开始测试指数平滑模型...")
+    test_exponential_smoothing_basic()
+    test_exponential_smoothing_simple()
+    test_exponential_smoothing_errors()
+    print("所有指数平滑测试通过!")