CLN: General clean of univariate volatility

Clean up types
Simplify cython directives

arch/univariate/base.py

@@ -22,7 +22,7 @@
 from arch.utility.exceptions import ConvergenceWarning, StartingValueWarning, \
     convergence_warning, starting_value_warning
 
-__all__ = ['implicit_constant', 'ARCHModelResult', 'ARCHModel']
+__all__ = ['implicit_constant', 'ARCHModelResult', 'ARCHModel', 'ARCHModelForecast']
 
 # Callback variables
 _callback_iter, _callback_llf = 0, 0.0,
@@ -583,7 +583,7 @@ def starting_values(self):
     @cached_property
     def num_params(self):
         """
-        Returns the number of parameters
+        Number of parameters in the model
         """
         raise NotImplementedError('Subclasses must implement')
 
@@ -658,7 +658,7 @@ def forecast(self, params, horizon=1, start=None, align='origin', method='analyt
 
         Parameters
         ----------
-        params : ndarray, optional
+        params : {ndarray, Series}, optional
             Alternative parameters to use.  If not provided, the parameters
             estimated when fitting the model are used.  Must be identical in
             shape to the parameters computed by fitting the model.
@@ -784,21 +784,8 @@ class ARCHModelFixedResult(_SummaryRepr):
     ----------
     loglikelihood : float
         Value of the log-likelihood
-    aic : float
-        Akaike information criteria
-    bic : float
-        Schwarz/Bayes information criteria
-    conditional_volatility : {ndarray, Series}
-        nobs element array containing the conditional volatility (square root
-        of conditional variance).  The values are aligned with the input data
-        so that the value in the t-th position is the variance of t-th error,
-        which is computed using time-(t-1) information.
     params : Series
         Estimated parameters
-    nobs : int
-        Number of observations used in the estimation
-    num_params : int
-        Number of parameters in the model
     resid : {ndarray, Series}
         nobs element array containing model residuals
     model : ARCHModel
@@ -919,9 +906,9 @@ def summary(self):
                                 headers=header, title=title)
             smry.tables.append(table)
 
-        extra_text = ('Results generated with user-specified parameters.',
+        extra_text = ['Results generated with user-specified parameters.',
                       'Since the model was not estimated, there are no std. '
-                      'errors.')
+                      'errors.']
         smry.add_extra_txt(extra_text)
         return smry
 
@@ -960,6 +947,14 @@ def params(self):
     def conditional_volatility(self):
         """
         Estimated conditional volatility
+
+        Returns
+        -------
+        conditional_volatility : {ndarray, Series}
+            nobs element array containing the conditional volatility (square
+            root of conditional variance).  The values are aligned with the
+            input data so that the value in the t-th position is the variance
+            of t-th error, which is computed using time-(t-1) information.
         """
         if self._is_pandas:
             return pd.Series(self._volatility,
@@ -971,7 +966,7 @@ def conditional_volatility(self):
     @cached_property
     def nobs(self):
         """
-        Number of data points used ot estimate model
+        Number of data points used to estimate model
         """
         return self._nobs
 
@@ -1274,33 +1269,10 @@ class ARCHModelResult(ARCHModelFixedResult):
     ----------
     loglikelihood : float
         Value of the log-likelihood
-    aic : float
-        Akaike information criteria
-    bic : float
-        Schwarz/Bayes information criteria
-    conditional_volatility : {ndarray, Series}
-        nobs element array containing the conditional volatility (square root
-        of conditional variance).  The values are aligned with the input data
-        so that the value in the t-th position is the variance of t-th error,
-        which is computed using time-(t-1) information.
     params : Series
         Estimated parameters
     param_cov : DataFrame
         Estimated variance-covariance of the parameters
-    rsquared : float
-        R-squared
-    rsquared_adj : float
-        Degree of freedom adjusted R-squared
-    nobs : int
-        Number of observations used in the estimation
-    num_params : int
-        Number of parameters in the model
-    tvalues : Series
-        Array of t-statistics for the null the coefficient is 0
-    std_err : Series
-        Array of parameter standard errors
-    pvalues : Series
-        Array of p-values for the t-statistics
     resid : {ndarray, Series}
         nobs element array containing model residuals
     model : ARCHModel
@@ -1509,15 +1481,15 @@ def pvalues(self):
     @cached_property
     def std_err(self):
         """
-        Parameter standard error
+        Array of parameter standard errors
         """
         return pd.Series(np.sqrt(np.diag(self.param_cov)),
                          index=self._names, name='std_err')
 
     @cached_property
     def tvalues(self):
         """
-        t-statistics for the null the coefficient is 0
+        Array of t-statistics testing the null that the coefficient are 0
         """
         tvalues = self.params / self.std_err
         tvalues.name = 'tvalues'
@@ -1621,8 +1593,6 @@ class ARCHModelForecast(object):
         Forecast values for the conditional variance of the process
     residual_variance : DataFrame
         Forecast values for the conditional variance of the residuals
-    simulations : ARCHModelForecastSimulation
-        Object containing detailed simulation results if using a simulation-based method
     """
 
     def __init__(self, index, mean, variance, residual_variance,
@@ -1656,4 +1626,7 @@ def residual_variance(self):
 
     @property
     def simulations(self):
+        """
+        Detailed simulation results if using a simulation-based method
+        """
         return self._sim

arch/univariate/distribution.py

@@ -510,7 +510,8 @@ def __const_b(self, parameters):
         a = self.__const_a(parameters)
         return (1 + 3 * lam ** 2 - a ** 2) ** .5
 
-    def __const_c(self, parameters):
+    @staticmethod
+    def __const_c(parameters):
         """Compute c constant.
 
         Parameters
@@ -523,8 +524,8 @@ def __const_c(self, parameters):
         c : float
             Log of the constant used in loglikelihood
         """
-        eta, lam = parameters
-        #        return gamma((eta+1)/2) / ((pi*(eta-2))**.5 * gamma(eta/2))
+        eta = parameters[0]
+        # return gamma((eta+1)/2) / ((pi*(eta-2))**.5 * gamma(eta/2))
         return gammaln((eta + 1) / 2) - gammaln(eta / 2) - log(pi * (eta - 2)) / 2
 
     def ppf(self, arg, parameters):

arch/univariate/mean.py

@@ -97,7 +97,7 @@ class HARX(ARCHModel):
         nobs element vector containing the dependent variable
     x : {ndarray, DataFrame}, optional
         nobs by k element array containing exogenous regressors
-    lags : {scalar, array}, optional
+    lags : {scalar, ndarray}, optional
         Description of lag structure of the HAR.  Scalar included all lags
         between 1 and the value.  A 1-d array includes the HAR lags 1:lags[0],
         1:lags[1], ... A 2-d array includes the HAR lags of the form
@@ -199,13 +199,11 @@ def _static_gaussian_loglikelihood(resids):
     def _model_description(self, include_lags=True):
         """Generates the model description for use by __str__ and related
         functions"""
-        if include_lags:
-            if self.lags is not None:
-                lagstr = ['[' + str(lag[0]) + ':' + str(lag[1]) + ']'
-                          for lag in self._lags.T]
-                lagstr = ', '.join(lagstr)
-            else:
-                lagstr = 'none'
+        lagstr = 'none'
+        if include_lags and self.lags is not None:
+            lagstr = ['[' + str(lag[0]) + ':' + str(lag[1]) + ']'
+                      for lag in self._lags.T]
+            lagstr = ', '.join(lagstr)
         xstr = str(self._x.shape[1]) if self._x is not None else '0'
         conststr = 'yes' if self.constant else 'no'
         od = OrderedDict()
@@ -282,7 +280,7 @@ def simulate(self, params, nobs, burn=500, initial_value=None, x=None,
         initial_value : {ndarray, float}, optional
             Either a scalar value or `max(lags)` array set of initial values to
             use when initializing the model.  If omitted, 0.0 is used.
-        x : array, optional
+        x : {ndarray, DataFrame}, optional
             nobs + burn by k array of exogenous variables to include in the
             simulation.
         initial_value_vol : {ndarray, float}, optional
@@ -886,7 +884,7 @@ def simulate(self, params, nobs, burn=500, initial_value=None, x=None,
 
         Parameters
         ----------
-        params : array
+        params : {ndarray, DataFrame}
             Parameters to use when simulating the model.  Parameter order is
             [volatility distribution]. There are no mean parameters.
         nobs : int
@@ -1020,17 +1018,17 @@ def __init__(self, y=None, x=None, lags=None, constant=True,
     def _model_description(self, include_lags=True):
         """Generates the model description for use by __str__ and related
         functions"""
-        if include_lags:
-            if self.lags is not None:
-                lagstr = [str(lag[1]) for lag in self._lags.T]
-                lagstr = ', '.join(lagstr)
-            else:
-                lagstr = 'none'
+        lagstr = 'none'
+        if include_lags and self.lags is not None:
+            lagstr = [str(lag[1]) for lag in self._lags.T]
+            lagstr = ', '.join(lagstr)
+
         xstr = str(self._x.shape[1]) if self._x is not None else '0'
         conststr = 'yes' if self.constant else 'no'
         od = OrderedDict()
         od['constant'] = conststr
-        od['lags'] = lagstr
+        if include_lags:
+            od['lags'] = lagstr
         od['no. of exog'] = xstr
         od['volatility'] = self.volatility.__str__()
         od['distribution'] = self.distribution.__str__()

arch/univariate/recursions.pyx

@@ -1,6 +1,8 @@
+#!python
+#cython: language_level=3, boundscheck=False, wraparound=False, cdivision=True
+
 import numpy as np
 cimport numpy as np
-cimport cython
 
 __all__ = ['harch_recursion','arch_recursion','garch_recursion',
            'egarch_recursion']
@@ -16,9 +18,6 @@ cdef extern from 'float.h':
 
 cdef double LNSIGMA_MAX = log(DBL_MAX)
 
-@cython.boundscheck(False)
-@cython.wraparound(False)
-@cython.cdivision(True)
 def harch_recursion(double[::1] parameters,
                     double[::1] resids,
                     double[::1] sigma2,
@@ -69,9 +68,6 @@ def harch_recursion(double[::1] parameters,
 
     return np.asarray(sigma2)
 
-@cython.boundscheck(False)
-@cython.wraparound(False)
-@cython.cdivision(True)
 def arch_recursion(double[::1] parameters,
                    double[::1] resids,
                    double[::1] sigma2,
@@ -121,9 +117,6 @@ def arch_recursion(double[::1] parameters,
 
     return np.asarray(sigma2)
 
-@cython.boundscheck(False)
-@cython.wraparound(False)
-@cython.cdivision(True)
 def garch_recursion(double[::1] parameters,
                     double[::1] fresids,
                     double[::1] sresids,
@@ -199,9 +192,6 @@ def garch_recursion(double[::1] parameters,
 
     return np.asarray(sigma2)
 
-@cython.boundscheck(False)
-@cython.wraparound(False)
-@cython.cdivision(True)
 def egarch_recursion(double[::1] parameters,
                      double[::1] resids,
                      double[::1] sigma2,
@@ -282,9 +272,6 @@ def egarch_recursion(double[::1] parameters,
 
 
 
-@cython.boundscheck(False)
-@cython.wraparound(False)
-@cython.cdivision(True)
 def midas_recursion(double[::1] parameters,
                     double[::1] weights,
                     double[::1] resids,

arch/univariate/volatility.py

@@ -25,7 +25,7 @@
                                                    egarch_recursion, midas_recursion)
 
 __all__ = ['GARCH', 'ARCH', 'HARCH', 'ConstantVariance', 'EWMAVariance', 'RiskMetrics2006',
-           'EGARCH', 'FixedVariance', 'BootstrapRng', 'MIDASHyperbolic']
+           'EGARCH', 'FixedVariance', 'BootstrapRng', 'MIDASHyperbolic', 'VolatilityProcess']
 
 
 class BootstrapRng(object):

arch/utility/array.py

@@ -12,7 +12,7 @@
 from pandas import DataFrame, NaT, Series, Timestamp, to_datetime
 
 __all__ = ['ensure1d', 'parse_dataframe', 'DocStringInheritor',
-           'date_to_index']
+           'date_to_index', 'cutoff_to_index']
 
 deprecation_doc = """
 {func} has been moved.  Please use {new_location}.{func}.