mixture model

prefsampling/approval/__init__.py

@@ -14,6 +14,10 @@
 from prefsampling.approval.euclidean import euclidean
 from prefsampling.approval.partylist import urn_partylist
 from prefsampling.approval.truncated_urn import truncated_urn
+from prefsampling.approval.mixture import (
+    mixture,
+    concat,
+)
 
 
 __all__ = [
@@ -29,4 +33,6 @@
     "euclidean",
     "urn_partylist",
     "truncated_urn",
+    "mixture",
+    "concat"
 ]

prefsampling/approval/mixture.py

@@ -0,0 +1,81 @@
+import numpy as np
+
+
+def mixture(
+        num_voters: int,
+        num_candidates: int,
+        weights: list,
+        list_of_cultures: list,
+        list_of_params: list,
+        seed: int = None,
+) -> list[set[int]]:
+    """
+    Generates a mixture of approval votes from different cultures.
+
+    Parameters
+    ----------
+        num_voters : int
+            Number of Voters.
+        num_candidates : int
+            Number of Candidates.
+        weights: list
+            Probability of being sampled from a given culture
+        list_of_cultures: list,
+            List of the cultures.
+        list_of_params: list,
+            List of params for each considered culture
+            (excluding the number of voters and the number of candidates).
+        seed : int
+            Seed for numpy random number generator.
+            Note that this is only the seed for this function.
+            If you want to use particular seed for the functions generating votes,
+            you should pass it as argument within the list of params.
+
+    Returns
+    -------
+        list[set[int]]
+            Approval votes.
+    """
+    rng = np.random.default_rng(seed)
+
+    options = [i for i in range(len(weights))]
+    sum_weights = sum(weights)
+    norm_weights = [w / sum_weights for w in weights]
+    samples = list(rng.choice(options, size=num_voters, replace=True, p=norm_weights))
+    list_of_num_voters = [samples.count(i) for i in range(len(weights))]
+    return concat(list_of_num_voters, num_candidates, list_of_cultures, list_of_params)
+
+
+def concat(
+        list_of_num_voters: list,
+        num_candidates: int,
+        list_of_cultures: list,
+        list_of_params: list
+) -> list[set[int]]:
+    """
+    Generates a concatenation of approval votes from different cultures.
+
+    Parameters
+    ----------
+        list_of_num_voters : int
+            List of numbers of voters to be sampled from each culture.
+        num_candidates : int
+            Number of Candidates.
+        list_of_cultures: list,
+            List of the cultures.
+        list_of_params: list,
+            List of params for each considered culture
+            (excluding the number of voters and the number of candidates).
+
+    Returns
+    -------
+        list[set[int]]
+            Approval votes.
+    """
+    all_votes = []
+    for num_voters, culture, params in zip(list_of_num_voters, list_of_cultures, list_of_params):
+        if num_voters > 0:
+            votes = culture(num_voters, num_candidates, **params)
+            for vote in votes:
+                all_votes.append(vote)
+    return all_votes

prefsampling/ordinal/__init__.py

@@ -28,6 +28,10 @@
 from prefsampling.ordinal.identity import identity
 from prefsampling.ordinal.didi import didi
 from prefsampling.ordinal.identity import identity
+from prefsampling.ordinal.mixture import (
+    mixture,
+    concat,
+)
 
 __all__ = [
     "urn",
@@ -47,4 +51,6 @@
     "TreeSampler",
     "identity",
     "didi",
+    "mixture",
+    "concat",
 ]

prefsampling/ordinal/mixture.py

@@ -0,0 +1,80 @@
+import numpy as np
+
+
+def mixture(
+        num_voters: int,
+        num_candidates: int,
+        weights: list,
+        list_of_cultures: list,
+        list_of_params: list,
+        seed: int = None,
+) -> np.ndarray:
+    """
+    Generates a mixture of ordinal votes from different cultures.
+
+    Parameters
+    ----------
+        num_voters : int
+            Number of Voters.
+        num_candidates : int
+            Number of Candidates.
+        weights: list
+            Probability of being sampled from a given culture
+        list_of_cultures: list,
+            List of the cultures.
+        list_of_params: list,
+            List of params for each considered culture
+            (excluding the number of voters and the number of candidates).
+        seed : int
+            Seed for numpy random number generator.
+            Note that this is only the seed for this function.
+            If you want to use particular seed for the functions generating votes,
+            you should pass it as argument within the list of params.
+
+    Returns
+    -------
+        np.ndarray
+            Ordinal votes.
+    """
+    rng = np.random.default_rng(seed)
+
+    options = [i for i in range(len(weights))]
+    sum_weights = sum(weights)
+    norm_weights = [w / sum_weights for w in weights]
+    samples = list(rng.choice(options, size=num_voters, replace=True, p=norm_weights))
+    list_of_num_voters = [samples.count(i) for i in range(len(weights))]
+    return concat(list_of_num_voters, num_candidates, list_of_cultures, list_of_params)
+
+
+def concat(
+        list_of_num_voters: list,
+        num_candidates: int,
+        list_of_cultures: list,
+        list_of_params: list
+) -> np.ndarray:
+    """
+    Generates a concatenation of ordinal votes from different cultures.
+
+    Parameters
+    ----------
+        list_of_num_voters : int
+            List of numbers of voters to be sampled from each culture.
+        num_candidates : int
+            Number of Candidates.
+        list_of_cultures: list,
+            List of the cultures.
+        list_of_params: list,
+            List of params for each considered culture
+            (excluding the number of voters and the number of candidates).
+
+    Returns
+    -------
+        np.ndarray
+            Ordinal votes.
+    """
+    all_votes = np.empty(shape=(0, num_candidates), dtype=int)
+    for num_voters, culture, params in zip(list_of_num_voters, list_of_cultures, list_of_params):
+        if num_voters > 0:
+            votes = culture(num_voters, num_candidates, **params)
+            all_votes = np.concatenate((all_votes, votes), axis=0)
+    return all_votes