aigroup-econ-mcp 1.4.3__py3-none-any.whl → 2.0.1__py3-none-any.whl

econometrics/specific_data_modeling/micro_discrete_limited_data/__init__.py

@@ -0,0 +1,40 @@
+"""
+微观离散与受限数据模型模块
+"""
+
+# 离散选择模型
+from .discrete_choice_models import (
+    LogitModel,
+    ProbitModel,
+    MultinomialLogit,
+    OrderedLogit,
+    ConditionalLogit
+)
+
+# 受限因变量模型
+from .limited_dependent_variable_models import (
+    TobitModel,
+    HeckmanModel
+)
+
+# 计数数据模型
+from .count_data_models import (
+    PoissonModel,
+    NegativeBinomialModel,
+    ZeroInflatedPoissonModel,
+    ZeroInflatedNegativeBinomialModel
+)
+
+__all__ = [
+    'LogitModel',
+    'ProbitModel',
+    'MultinomialLogit',
+    'OrderedLogit',
+    'ConditionalLogit',
+    'TobitModel',
+    'HeckmanModel',
+    'PoissonModel',
+    'NegativeBinomialModel',
+    'ZeroInflatedPoissonModel',
+    'ZeroInflatedNegativeBinomialModel'
+]

econometrics/specific_data_modeling/micro_discrete_limited_data/count_data_models.py

@@ -0,0 +1,311 @@
+"""
+计数数据模型模块
+基于statsmodels等现有库实现
+"""
+
+import numpy as np
+import pandas as pd
+import math
+from scipy import stats
+try:
+    import statsmodels.api as sm
+    from statsmodels.discrete.discrete_model import Poisson, NegativeBinomial
+    from statsmodels.discrete.count_model import ZeroInflatedPoisson, ZeroInflatedNegativeBinomialP
+    HAS_STATSMODELS = True
+except ImportError:
+    HAS_STATSMODELS = False
+    Poisson = NegativeBinomial = ZeroInflatedPoisson = ZeroInflatedNegativeBinomialP = None
+
+
+class _PlaceholderModel:
+    def __init__(self, *args, **kwargs):
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+    
+    def fit(self, *args, **kwargs):
+        pass
+
+
+class PoissonModel:
+    """
+    泊松回归模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self):
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合泊松回归模型"""
+        X = np.array(X)
+        y = np.array(y)
+        
+        if np.any(y < 0) or np.any(y != np.floor(y)):
+            raise ValueError("因变量必须是非负整数")
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 拟合模型
+        self.model_ = Poisson(y, X_with_const)
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict(self, X):
+        """预测计数期望值"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict_proba(self, X, max_count=20):
+        """预测计数概率分布"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        # 使用statsmodels的预测方法
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        mu = self.results_.predict(X_with_const)
+        
+        # 计算泊松概率
+        probas = []
+        for k in range(max_count + 1):
+            prob = np.exp(-mu) * (mu ** k) / math.factorial(k)
+            probas.append(prob)
+        
+        return np.array(probas).T
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class NegativeBinomialModel:
+    """
+    负二项回归模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self, distr='nb2'):
+        """
+        初始化负二项回归模型
+        
+        参数:
+        distr: 分布类型，'nb1' 或 'nb2' (默认)
+        """
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+        self.distr = distr
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合负二项回归模型"""
+        X = np.array(X)
+        y = np.array(y)
+        
+        if np.any(y < 0) or np.any(y != np.floor(y)):
+            raise ValueError("因变量必须是非负整数")
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 拟合模型
+        self.model_ = NegativeBinomial(y, X_with_const, loglike_method=self.distr)
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict(self, X):
+        """预测计数期望值"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict_proba(self, X, max_count=20):
+        """预测计数概率分布"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        # 使用模型预测均值
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        mu = self.results_.predict(X_with_const)
+        
+        # 获取alpha参数
+        alpha = self.results_.params[-1]  # 最后一个参数是ln(alpha)
+        alpha = np.exp(alpha)
+        
+        # 计算负二项概率 (NB2参数化)
+        probas = []
+        for k in range(max_count + 1):
+            # 负二项概率质量函数 - 使用scipy.stats
+            prob = stats.nbinom.pmf(k, 1/alpha, 1/(1 + alpha * mu))
+            probas.append(prob)
+        
+        return np.array(probas).T
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class ZeroInflatedPoissonModel:
+    """
+    零膨胀泊松模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self, exog_infl=None):
+        """
+        初始化零膨胀泊松模型
+        
+        参数:
+        exog_infl: 用于零膨胀部分的解释变量，默认为None（使用与计数部分相同的变量）
+        """
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+        self.exog_infl = exog_infl
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合零膨胀泊松模型"""
+        X = np.array(X)
+        y = np.array(y)
+        
+        if np.any(y < 0) or np.any(y != np.floor(y)):
+            raise ValueError("因变量必须是非负整数")
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 零膨胀部分的解释变量
+        if self.exog_infl is not None:
+            exog_infl = sm.add_constant(np.array(self.exog_infl))
+        else:
+            exog_infl = X_with_const
+        
+        # 拟合模型
+        self.model_ = ZeroInflatedPoisson(
+            endog=y, 
+            exog=X_with_const, 
+            exog_infl=exog_infl,
+            inflation='logit'
+        )
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict(self, X):
+        """预测计数期望值"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict_proba(self, X):
+        """预测计数概率分布"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(which='prob', exog=X_with_const)
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class ZeroInflatedNegativeBinomialModel:
+    """
+    零膨胀负二项模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self, exog_infl=None, distr='nb2'):
+        """
+        初始化零膨胀负二项模型
+        
+        参数:
+        exog_infl: 用于零膨胀部分的解释变量，默认为None（使用与计数部分相同的变量）
+        distr: 分布类型，'nb1' 或 'nb2' (默认)
+        """
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+        self.exog_infl = exog_infl
+        self.distr = distr
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合零膨胀负二项模型"""
+        X = np.array(X)
+        y = np.array(y)
+        
+        if np.any(y < 0) or np.any(y != np.floor(y)):
+            raise ValueError("因变量必须是非负整数")
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 零膨胀部分的解释变量
+        if self.exog_infl is not None:
+            exog_infl = sm.add_constant(np.array(self.exog_infl))
+        else:
+            exog_infl = X_with_const
+        
+        # 拟合模型
+        self.model_ = ZeroInflatedNegativeBinomialP(
+            endog=y, 
+            exog=X_with_const, 
+            exog_infl=exog_infl,
+            inflation='logit',
+            loglike_method=self.distr
+        )
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict(self, X):
+        """预测计数期望值"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict_proba(self, X):
+        """预测计数概率分布"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(which='prob', exog=X_with_const)
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+# 如果statsmodels不可用，则使用占位符
+if not HAS_STATSMODELS:
+    PoissonModel = _PlaceholderModel
+    NegativeBinomialModel = _PlaceholderModel
+    ZeroInflatedPoissonModel = _PlaceholderModel
+    ZeroInflatedNegativeBinomialModel = _PlaceholderModel

econometrics/specific_data_modeling/micro_discrete_limited_data/discrete_choice_models.py

@@ -0,0 +1,294 @@
+"""
+离散选择模型模块
+基于statsmodels等现有库实现
+"""
+
+import numpy as np
+import pandas as pd
+try:
+    import statsmodels.api as sm
+    from statsmodels.discrete.discrete_model import Logit, Probit, MNLogit
+    from statsmodels.miscmodels.ordinal_model import OrderedModel
+    # 注意: statsmodels目前没有内置ConditionalLogit，需要自定义或使用其他库
+    HAS_STATSMODELS = True
+except ImportError:
+    HAS_STATSMODELS = False
+    Logit = Probit = MNLogit = OrderedModel = None
+
+# 占位符类以防statsmodels不可用
+class _PlaceholderModel:
+    def __init__(self, *args, **kwargs):
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+    
+    def fit(self, *args, **kwargs):
+        pass
+
+
+class LogitModel:
+    """
+    Logistic回归模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self):
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合Logistic回归模型"""
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+            
+        X = np.array(X)
+        y = np.array(y)
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 拟合模型
+        self.model_ = Logit(y, X_with_const)
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict_proba(self, X):
+        """预测概率"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict(self, X, threshold=0.5):
+        """预测类别"""
+        proba = self.predict_proba(X)
+        return (proba >= threshold).astype(int)
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class ProbitModel:
+    """
+    Probit回归模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self):
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y):
+        """拟合Probit回归模型"""
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+            
+        X = np.array(X)
+        y = np.array(y)
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 拟合模型
+        self.model_ = Probit(y, X_with_const)
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict_proba(self, X):
+        """预测概率"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict(self, X, threshold=0.5):
+        """预测类别"""
+        proba = self.predict_proba(X)
+        return (proba >= threshold).astype(int)
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class MultinomialLogit:
+    """
+    多项Logit模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self):
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        self.classes_ = None
+        
+    def fit(self, X, y):
+        """拟合多项Logit模型"""
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+            
+        X = np.array(X)
+        y = np.array(y)
+        
+        self.classes_ = np.unique(y)
+        
+        # 添加常数项
+        X_with_const = sm.add_constant(X)
+        
+        # 拟合模型
+        self.model_ = MNLogit(y, X_with_const)
+        self.results_ = self.model_.fit(disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict_proba(self, X):
+        """预测各类别的概率"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        X_with_const = sm.add_constant(X)
+        return self.results_.predict(X_with_const)
+    
+    def predict(self, X):
+        """预测类别"""
+        proba = self.predict_proba(X)
+        return self.classes_[np.argmax(proba, axis=1)]
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class OrderedLogit:
+    """
+    有序Logit模型 (基于statsmodels实现)
+    """
+    
+    def __init__(self):
+        self.model_ = None
+        self.results_ = None
+        self.fitted_ = False
+        self.classes_ = None
+        
+    def fit(self, X, y):
+        """拟合有序Logit模型"""
+        if not HAS_STATSMODELS:
+            raise ImportError("需要安装statsmodels库: pip install statsmodels")
+            
+        X = np.array(X)
+        y = np.array(y)
+        
+        self.classes_ = np.unique(y)
+        
+        # OrderedModel不允许包含常数项
+        # 直接使用X，不添加常数项
+        self.model_ = OrderedModel(y, X, distr='logit')
+        self.results_ = self.model_.fit(method='bfgs', disp=0)
+        self.fitted_ = True
+        return self
+    
+    def predict_proba(self, X):
+        """预测各类别的概率"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        # 预测时也不添加常数项
+        return self.results_.predict(X)
+    
+    def predict(self, X):
+        """预测类别"""
+        proba = self.predict_proba(X)
+        return self.classes_[np.argmax(proba, axis=1)]
+    
+    def summary(self):
+        """返回模型摘要"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        return self.results_.summary()
+
+
+class ConditionalLogit:
+    """
+    条件Logit模型 
+    注意: statsmodels目前没有内置实现，此为简化版本
+    """
+    
+    def __init__(self):
+        self.params_ = None
+        self.fitted_ = False
+        
+    def fit(self, X, y, groups):
+        """拟合条件Logit模型"""
+        # 简化的条件Logit实现
+        # 在实际应用中可能需要使用其他专门库如pylogit或mne-logit
+        X = np.array(X)
+        y = np.array(y)
+        groups = np.array(groups)
+        
+        n_samples, n_features = X.shape
+        
+        # 使用scipy优化器进行简单实现
+        from scipy.optimize import minimize
+        
+        def neg_log_likelihood(params):
+            beta = params
+            loglik = 0
+            unique_groups = np.unique(groups)
+            
+            for group_id in unique_groups:
+                group_mask = (groups == group_id)
+                X_g = X[group_mask]
+                y_g = y[group_mask]
+                
+                scores = np.dot(X_g, beta)
+                probs = np.exp(scores)
+                probs = probs / np.sum(probs)
+                probs = np.clip(probs, 1e-15, 1-1e-15)
+                
+                loglik += np.sum(y_g * np.log(probs))
+            
+            return -loglik
+        
+        initial_params = np.random.normal(0, 0.1, n_features)
+        result = minimize(neg_log_likelihood, initial_params, method='BFGS')
+        
+        if result.success:
+            self.params_ = result.x
+            self.fitted_ = True
+        else:
+            raise RuntimeError("模型优化失败")
+            
+        return self
+    
+    def predict_proba(self, X):
+        """预测概率"""
+        if not self.fitted_:
+            raise ValueError("模型尚未拟合")
+        X = np.array(X)
+        scores = np.dot(X, self.params_)
+        exp_scores = np.exp(scores - np.max(scores))  # 数值稳定性
+        return exp_scores / np.sum(exp_scores)
+    
+    def predict(self, X):
+        """预测类别"""
+        proba = self.predict_proba(X)
+        return (proba >= 0.5).astype(int)
+
+
+# 如果statsmodels不可用，则使用占位符
+if not HAS_STATSMODELS:
+    LogitModel = _PlaceholderModel
+    ProbitModel = _PlaceholderModel
+    MultinomialLogit = _PlaceholderModel
+    OrderedLogit = _PlaceholderModel
+    ConditionalLogit = _PlaceholderModel