Develop batch prediction

bayesml/bernoulli/_bernoulli.py

@@ -431,10 +431,7 @@ def calc_pred_dist(self):
         return self
 
     def _calc_pred_density(self,x):
-        if x:
-            return self.p_theta
-        else:
-            return 1.0-self.p_theta
+        return np.where(x==1,self.p_theta,1.0-self.p_theta)
 
     def make_prediction(self,loss="squared"):
         """Predict a new data point under the given criterion.

bayesml/linearregression/_linearregression.py

@@ -326,12 +326,12 @@ class LearnModel(base.Posterior,base.PredictiveMixin):
         a positive real number
     hn_beta : float
         a positive real number
-    p_m : float
-        a positive real number
-    p_lambda : float
-        a positive real number
-    p_nu : float
-        a positive real number
+    p_ms : numpy ndarray
+        positive real numbers
+    p_lambdas : numpy ndarray
+        positive real numbers
+    p_nus : numpy ndarray
+        positive real numbers
     """
     def __init__(
             self,
@@ -357,9 +357,9 @@ def __init__(
         self.hn_beta = 1.0
 
         # p_params
-        self.p_m = 0.0
-        self.p_lambda =  1.0
-        self.p_nu = 2.0
+        self.p_ms = 0.0
+        self.p_lambdas =  1.0
+        self.p_nus = 2.0
 
         # sample size
         self._n = 0
@@ -510,9 +510,10 @@ def _check_sample_x(self,x):
         _check.float_vecs(x,'x',DataFormatError)
         if x.shape[-1] != self.c_degree:
             raise(DataFormatError(f"x.shape[-1] must be c_degree:{self.c_degree}"))
-
+        return x.reshape(-1,self.c_degree)
+
     def _check_sample_y(self,y):
-        _check.floats(y,'y',DataFormatError)
+        return _check.floats(y,'y',DataFormatError)
 
     def _check_sample(self,x,y):
         self._check_sample_x(x)
@@ -696,38 +697,37 @@ def get_p_params(self):
 
         Returns
         -------
-        p_params : dict of {str: float}
-            * ``"p_m"`` : The value of ``self.p_m``
-            * ``"p_lambda"`` : The value of ``self.p_lambda``
-            * ``"p_nu"`` : The value of ``self.p_nu``
+        p_params : dict of {str: numpy ndarray}
+            * ``"p_ms"`` : The value of ``self.p_ms``
+            * ``"p_lambdas"`` : The value of ``self.p_lambdas``
+            * ``"p_nus"`` : The value of ``self.p_nus``
         """
-        return {"p_m":self.p_m, "p_lambda":self.p_lambda, "p_nu":self.p_nu}
+        return {"p_ms":self.p_ms, "p_lambdas":self.p_lambdas, "p_nus":self.p_nus}
 
     def calc_pred_dist(self, x):
         """Calculate the parameters of the predictive distribution.
 
         Parameters
         ----------
         x : numpy ndarray
-            1 dimensional float array whose size is ``self.c_degree``
+            float array. The size along the last dimension must conincides with the c_degree.
+            If you want to use a constant term, it should be included in x.
         """
-        _check.float_vec(x,'x',DataFormatError)
-        if x.shape != (self.c_degree,):
-            raise(DataFormatError("x must be a 1 dimensional float array whose size coincide with ``self.c_degree``"))
-        self.p_m = x @ self.hn_mu_vec
-        self.p_lambda = self.hn_alpha / self.hn_beta / (1.0 + x @ np.linalg.solve(self.hn_lambda_mat,x))
-        self.p_nu = 2.0 * self.hn_alpha
+        x = self._check_sample_x(x)
+        self.p_ms = x @ self.hn_mu_vec
+        self.p_lambdas = self.hn_alpha / self.hn_beta / (1.0 + np.sum(x.T * np.linalg.solve(self.hn_lambda_mat,x.T),axis=0))
+        self.p_nus = np.ones(x.shape[0]) * 2.0 * self.hn_alpha
         return self
 
     def _calc_pred_dist(self, x):
         """Calculate predictive distribution without check."""
-        self.p_m = x @ self.hn_mu_vec
-        self.p_lambda = self.hn_alpha / self.hn_beta / (1.0 + x @ np.linalg.solve(self.hn_lambda_mat,x))
-        self.p_nu = 2.0 * self.hn_alpha
+        self.p_ms = x @ self.hn_mu_vec
+        self.p_lambdas = self.hn_alpha / self.hn_beta / (1.0 + np.sum(x.T * np.linalg.solve(self.hn_lambda_mat,x.T),axis=0))
+        self.p_nus = np.ones(x.shape[0]) * 2.0 * self.hn_alpha
         return self
 
     def _calc_pred_density(self,y):
-        return ss_t.pdf(y,loc=self.p_m, scale=1.0/np.sqrt(self.p_lambda), df=self.p_nu)
+        return ss_t.pdf(y,loc=self.p_ms, scale=1.0/np.sqrt(self.p_lambdas), df=self.p_nus)
 
     def make_prediction(self,loss="squared"):
         """Predict a new data point under the given criterion.
@@ -740,20 +740,21 @@ def make_prediction(self,loss="squared"):
 
         Returns
         -------
-        Predicted_value : {float, rv_frozen}
-            The predicted value under the given loss function. 
+        Predicted_values : {numpy ndarray, rv_frozen}
+            The predicted values under the given loss function. 
+            The size of the predicted values is the same as the sample size of x when you called calc_pred_dist(x).
             If the loss function is \"KL\", the predictive distribution itself will be returned
-            as rv_frozen object of scipy.stats.
+            as rv_frozen object of scipy.stats. The rv_frozen object supports broadcasting.
 
         See Also
         --------
         scipy.stats.rv_continuous
         scipy.stats.rv_discrete
         """
         if loss == "squared" or loss == "0-1" or loss == "abs":
-            return self.p_m
+            return self.p_ms
         elif loss == "KL":
-            return ss_t(loc=self.p_m, scale=1.0/np.sqrt(self.p_lambda), df=self.p_nu)
+            return ss_t(loc=self.p_ms, scale=1.0/np.sqrt(self.p_lambdas), df=self.p_nus)
         else:
             raise(CriteriaError("Unsupported loss function! "
                                 "This function supports \"squared\", \"0-1\", \"abs\", and \"KL\"."))
@@ -764,7 +765,8 @@ def pred_and_update(self,x,y,loss="squared"):
         Parameters
         ----------
         x : numpy ndarray
-            1 dimensional float array whose size is ``self.c_degree``.
+            float array. The size along the last dimension must conincides with the c_degree.
+            If you want to use a constant term, it should be included in x.
         y : float
 
         loss : str, optional
@@ -773,8 +775,9 @@ def pred_and_update(self,x,y,loss="squared"):
 
         Returns
         -------
-        Predicted_value : {float, rv_frozen}
-            The predicted value under the given loss function. 
+        Predicted_values : {numpy ndarray, rv_frozen}
+            The predicted values under the given loss function. 
+            The size of the predicted values is the same as the sample size of x when you called calc_pred_dist(x).
             If the loss function is \"KL\", the predictive distribution itself will be returned
             as rv_frozen object of scipy.stats.
         """
@@ -808,11 +811,12 @@ def calc_pred_var(self):
 
         Returns
         -------
-        var : float
-            The variance of the predictive distribution.
+        vars : numpy ndarray
+            The variances of the predictive distribution. 
+            The size of the vars is the same as the sample size of x when you called calc_pred_dist(x).
         """
-        if self.p_nu > 2:
-            return self.p_nu / self.p_lambda / (self.p_nu-2)
-        else:
-            warnings.warn("Variance of the predictive distribution cannot defined for the current p_nu.",ResultWarning)
-            return None
+        indices = self.p_nus > 2
+        var = np.empty(self.p_nus.shape[0])
+        var[indices] = self.p_nus[indices] / self.p_lambdas[indices] / (self.p_nus[indices]-2)
+        var[~indices] = np.nan
+        return var