rename variables in EnsembleDistribution consstructor, begin to imple…

…ment MSCABeta

src/ensemble/distributions.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from abc import ABC, abstractmethod
 from typing import Tuple, Union
 
@@ -17,9 +19,17 @@ class Distribution(ABC):
 
     """
 
-    def __init__(self, mean: float = None, variance: float = None):
+    def __init__(
+        self,
+        mean: float = None,
+        variance: float = None,
+        lb: float = None,
+        ub: float = None,
+    ):
         self.mean = mean
         self.variance = variance
+        self.lb = lb
+        self.ub = ub
         # # some kind of dictionary with
         # #   key: the support (full real line, semi infinite, etc...)
         # #   value: function that gets called when distribution is initialized
@@ -256,6 +266,13 @@ def _create_scipy_dist(self) -> None:
 
 # analytic sol
 class Beta(Distribution):
+    # TODO: WANT TO BE ABLE TO PASS IN UPPER AND LOWER BOUNDS TO BE REFLECTED IN THE DIST
+    # EX: MEAN 6, VAR 0.2, LB 5, UB 10
+    # ADJ_MEAN = (MEAN - LB) / INTERVAL_WIDTH
+    # ADJ_VAR = VAR / INTERVAL_WIDTH
+    # INPUT ADJ MEAN & VAR INTO FUNCTION
+    # JUST GET RVS TO WORK FOR NOW, WHEN YOU TAKE A SAMPLE OF SIZE 100,
+    # JUST MULTIPLICATIVELY SCALE AND THEN LINERALY SHIFT THE DATA TO THE ORIGINAL BOUNDS
     """https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.beta.html#scipy.stats.beta"""
 
     def support(self) -> Tuple[float, float]:
@@ -275,6 +292,87 @@ def _create_scipy_dist(self) -> None:
         self._scipy_dist = stats.beta(a=alpha, b=beta)
 
 
+class MSCABeta(Distribution):
+    def _create_scipy_dist(self) -> None:
+        self.width = self.ub - self.lb
+        adj_mean = (self.mean - self.lb) / self.width
+        adj_var = self.variance / self.width
+        self._scipy_dist = Beta(adj_mean, adj_var)
+
+    def support(self) -> Tuple[float, float]:
+        """create tuple representing endpoints of support"""
+
+    def rvs(self, *args, **kwds):
+        """defaults to scipy implementation for generating random variates
+
+        Returns
+        -------
+        np.ndarray
+            random variates from a given distribution/parameters
+        """
+        return (self._scipy_dist.rvs(*args, **kwds) + self.lb) * self.width
+
+    def pdf(self, x: npt.ArrayLike) -> np.ndarray:
+        """defaults to scipy implementation for probability density function
+
+        Parameters
+        ----------
+        x : npt.ArrayLike
+            quantiles
+
+        Returns
+        -------
+        np.ndarray
+            PDF evaluated at quantile x
+        """
+        return (self._scipy_dist.pdf(x) + self.lb) * self.width
+
+    # def cdf(self, q: npt.ArrayLike) -> np.ndarray:
+    #     """defaults to scipy implementation for cumulative density function
+
+    #     Parameters
+    #     ----------
+    #     q : npt.ArrayLike
+    #         quantiles
+
+    #     Returns
+    #     -------
+    #     np.ndarray
+    #         CDF evaluated at quantile q
+    #     """
+    #     return self._scipy_dist.cdf(q)
+
+    # def ppf(self, p: npt.ArrayLike) -> np.ndarray:
+    #     """defaults to scipy implementation for percent point function
+
+    #     Parameters
+    #     ----------
+    #     p : npt.ArrayLike
+    #         lower tail probability
+
+    #     Returns
+    #     -------
+    #     np.ndarray
+    #         PPF evaluated at lower tail probability p
+    #     """
+    #     return self._scipy_dist.ppf(p)
+
+    # def stats(self, moments: str) -> Union[float, Tuple[float, ...]]:
+    #     """defaults to scipy implementation for obtaining moments
+
+    #     Parameters
+    #     ----------
+    #     moments : str
+    #         m for mean, v for variance, s for skewness, k for kurtosis
+
+    #     Returns
+    #     -------
+    #     Union[float, Tuple[float, ...]]
+    #         mean, variance, skewness, and/or kurtosis
+    #     """
+    #     return self._scipy_dist.stats(moments=moments)
+
+
 distribution_dict = {
     "exponential": Exponential,
     "gamma": Gamma,
@@ -285,6 +383,7 @@ def _create_scipy_dist(self) -> None:
     "lognormal": LogNormal,
     "normal": Normal,
     "beta": Beta,
+    "MSCAbeta": MSCABeta,
 }
 
 

src/ensemble/model.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import json
 from typing import List, Tuple, Union
 
@@ -33,8 +35,6 @@ class EnsembleDistribution:
 
     def __init__(
         self,
-        # distributions: List[str],
-        # weights: List[float],
         named_weights: dict,
         mean: float,
         variance: float,
@@ -119,7 +119,9 @@ def pdf(self, x: npt.ArrayLike) -> np.ndarray:
         """
         return sum(
             weight * distribution.pdf(x)
-            for distribution, weight in zip(self.my_objs, self.weights)
+            for distribution, weight in zip(
+                self.fitted_distributions, self._weights
+            )
         )
 
     def cdf(self, q: npt.ArrayLike) -> np.ndarray:
@@ -138,7 +140,9 @@ def cdf(self, q: npt.ArrayLike) -> np.ndarray:
         """
         return sum(
             weight * distribution.cdf(q)
-            for distribution, weight in zip(self.my_objs, self.weights)
+            for distribution, weight in zip(
+                self.fitted_distributions, self._weights
+            )
         )
 
     def ppf(self, p: npt.ArrayLike) -> np.ndarray:
@@ -276,33 +280,61 @@ def to_json(self, file_path: str, appending: bool = False) -> None:
             with open(file_path, "w") as outfile:
                 json.dump([distribution_summary], outfile)
 
+    @classmethod
+    def from_json(cls, file_path: str) -> List[EnsembleDistribution]:
+        """deserializes JSON object into list of Ensemble Distribution objects
 
-def from_json(file_path: str) -> List[EnsembleDistribution]:
-    """deserializes JSON object into list of Ensemble Distribution objects
+        Parameters
+        ----------
+        file_path : str
+            path to file that JSON object is stored in
 
-    Parameters
-    ----------
-    file_path : str
-        path to file that JSON object is stored in
+        Returns
+        -------
+        List[EnsembleDistribution]
+            list of EnsembleDistribution objects
+        """
+        with open(file_path, "r") as infile:
+            distribution_summaries = json.load(infile)
+
+        res = [None] * len(distribution_summaries)
+        for i in range(len(distribution_summaries)):
+            named_weights, mean, variance = (
+                distribution_summaries[i]["named_weights"],
+                distribution_summaries[i]["mean"],
+                distribution_summaries[i]["variance"],
+            )
+            res[i] = cls(named_weights, mean, variance)
+
+        return res
 
-    Returns
-    -------
-    List[EnsembleDistribution]
-        list of EnsembleDistribution objects
-    """
-    with open(file_path, "r") as infile:
-        distribution_summaries = json.load(infile)
-
-    res = [None] * len(distribution_summaries)
-    for i in range(len(distribution_summaries)):
-        named_weights, mean, variance = (
-            distribution_summaries[i]["named_weights"],
-            distribution_summaries[i]["mean"],
-            distribution_summaries[i]["variance"],
-        )
-        res[i] = EnsembleDistribution(named_weights, mean, variance)
 
-    return res
+# def from_json(file_path: str) -> List[EnsembleDistribution]:
+#     """deserializes JSON object into list of Ensemble Distribution objects
+
+#     Parameters
+#     ----------
+#     file_path : str
+#         path to file that JSON object is stored in
+
+#     Returns
+#     -------
+#     List[EnsembleDistribution]
+#         list of EnsembleDistribution objects
+#     """
+#     with open(file_path, "r") as infile:
+#         distribution_summaries = json.load(infile)
+
+#     res = [None] * len(distribution_summaries)
+#     for i in range(len(distribution_summaries)):
+#         named_weights, mean, variance = (
+#             distribution_summaries[i]["named_weights"],
+#             distribution_summaries[i]["mean"],
+#             distribution_summaries[i]["variance"],
+#         )
+#         res[i] = EnsembleDistribution(named_weights, mean, variance)
+
+#     return res
 
 
 class EnsembleResult:

tests/test_distributions.py

@@ -94,6 +94,7 @@ def test_beta():
     beta = Beta(BETA_MEAN, BETA_VARIANCE)
     res = beta.stats(moments="mv")
     print("resulting mean and var: ", res)
+    assert False
     assert np.isclose(res[0], BETA_MEAN)
     assert np.isclose(res[1], BETA_VARIANCE)
 

tests/test_model.py

@@ -2,7 +2,7 @@
 import pytest
 import scipy.stats as stats
 
-from ensemble.model import EnsembleDistribution, EnsembleFitter, from_json
+from ensemble.model import EnsembleDistribution, EnsembleFitter
 
 STD_NORMAL_DRAWS = stats.norm(loc=0, scale=1).rvs(100)
 
@@ -78,7 +78,7 @@ def test_json():
     )
     model1.to_json("test_read.json", appending=True)
 
-    m1 = from_json("test_read.json")[1]
+    m1 = EnsembleDistribution.from_json("test_read.json")[1]
     assert m1.stats_temp("mv") == DEFAULT_SETTINGS
     assert m1._distributions == ["gamma", "invgamma"]
     assert m1._weights == [0.2, 0.8]