remove unnecesary calls to rv_frozen method (#134)

preliz/distributions/distributions.py

@@ -191,9 +191,9 @@ def _finite_endpoints(self, support):
             lower_ep, upper_ep = self.support
 
             if not np.isfinite(lower_ep) or support == "restricted":
-                lower_ep = self.rv_frozen.ppf(0.0001)
+                lower_ep = self.ppf(0.0001)
             if not np.isfinite(upper_ep) or support == "restricted":
-                upper_ep = self.rv_frozen.ppf(0.9999)
+                upper_ep = self.ppf(0.9999)
 
         return lower_ep, upper_ep
 

preliz/tests/test_plots.py

@@ -7,8 +7,8 @@
 
 
 @pytest.fixture(scope="function")
-def a_dist():
-    return pz.Beta(2, 6)
+def two_dist():
+    return pz.Beta(2, 6), pz.Poisson(4.5)
 
 
 @pytest.mark.parametrize(
@@ -24,8 +24,9 @@ def a_dist():
         {"ax": plt.subplots()[1]},
     ],
 )
-def test_plot_pdf_cdf_ppf(a_dist, kwargs):
-    a_dist.plot_pdf(**kwargs)
-    a_dist.plot_cdf(**kwargs)
-    kwargs.pop("support", None)
-    a_dist.plot_ppf(**kwargs)
+def test_continuous_plot_pdf_cdf_ppf(two_dist, kwargs):
+    for a_dist in two_dist:
+        a_dist.plot_pdf(**kwargs)
+        a_dist.plot_cdf(**kwargs)
+        kwargs.pop("support", None)
+        a_dist.plot_ppf(**kwargs)

preliz/utils/optimization.py

@@ -9,11 +9,10 @@
 def optimize_max_ent(dist, lower, upper, mass):
     def prob_bound(params, dist, lower, upper, mass):
         dist._update(*params)
-        rv_frozen = dist.rv_frozen
         if dist.kind == "discrete":
             lower -= 1
-        cdf0 = rv_frozen.cdf(lower)
-        cdf1 = rv_frozen.cdf(upper)
+        cdf0 = dist.cdf(lower)
+        cdf1 = dist.cdf(upper)
         loss = (cdf1 - cdf0) - mass
         return loss
 
@@ -47,7 +46,7 @@ def entropy_loss(params, dist):
 def optimize_quartile(dist, x_vals):
     def func(params, dist, x_vals):
         dist._update(*params)
-        loss = dist.rv_frozen.cdf(x_vals) - [0.25, 0.5, 0.75]
+        loss = dist.cdf(x_vals) - [0.25, 0.5, 0.75]
         return loss
 
     init_vals = dist.params
@@ -72,7 +71,7 @@ def func(params, dist, x_vals):
 def optimize_cdf(dist, x_vals, ecdf, **kwargs):
     def func(params, dist, x_vals, ecdf, **kwargs):
         dist._update(*params, **kwargs)
-        loss = dist.rv_frozen.cdf(x_vals) - ecdf
+        loss = dist.cdf(x_vals) - ecdf
         return loss
 
     init_vals = dist.params[:2]
@@ -115,7 +114,7 @@ def negll(params, dist, sample):
 def relative_error(dist, lower, upper, required_mass):
     if dist.kind == "discrete":
         lower -= 1
-    computed_mass = dist.rv_frozen.cdf(upper) - dist.rv_frozen.cdf(lower)
+    computed_mass = dist.cdf(upper) - dist.cdf(lower)
     return abs((computed_mass - required_mass) / required_mass * 100), computed_mass
 
 

preliz/utils/plot_utils.py

@@ -66,12 +66,12 @@ def plot_pdfpmf(dist, moments, pointinterval, quantiles, support, legend, figsiz
 
     x = dist.xvals(support)
     if dist.kind == "continuous":
-        density = dist.rv_frozen.pdf(x)
+        density = dist.pdf(x)
         ax.plot(x, density, label=label, color=color)
         ax.get_ylim()
         ax.set_yticks([])
     else:
-        mass = dist.rv_frozen.pmf(x)
+        mass = dist.pdf(x)
         eps = np.clip(dist._finite_endpoints(support), *dist.support)
         x_c = np.linspace(*eps, 1000)
 
@@ -111,7 +111,7 @@ def plot_cdf(dist, moments, pointinterval, quantiles, support, legend, figsize,
 
     eps = dist._finite_endpoints(support)
     x = np.linspace(*eps, 1000)
-    cdf = dist.rv_frozen.cdf(x)
+    cdf = dist.cdf(x)
     ax.plot(x, cdf, label=label, color=color)
 
     if pointinterval:
@@ -139,7 +139,7 @@ def plot_ppf(dist, moments, pointinterval, quantiles, legend, figsize, ax):
         label = None
 
     x = np.linspace(0, 1, 1000)
-    ax.plot(x, dist.rv_frozen.ppf(x), label=label, color=color)
+    ax.plot(x, dist.ppf(x), label=label, color=color)
 
     if pointinterval:
         plot_pointinterval(dist.rv_frozen, quantiles=quantiles, rotated=True, ax=ax)