Merge pull request #308 from data-8/multinomial

add sample_proportions

Author: SamLau95

README.md

@@ -27,9 +27,13 @@ pip install datascience
 
 This project adheres to [Semantic Versioning](http://semver.org/).
 
+### v0.10.4
+
+- Add `sample_proportions` function.
+
 ### v0.10.3
 
-- Fix `OrderedDict` bug in `Table.hist`
+- Fix `OrderedDict` bug in `Table.hist`.
 
 ### v0.10.2
 

datascience/util.py

@@ -1,7 +1,8 @@
 """Utility functions"""
 
 __all__ = ['make_array', 'percentile', 'plot_cdf_area', 'plot_normal_cdf',
-           'table_apply', 'proportions_from_distribution', 'minimize']
+           'table_apply', 'proportions_from_distribution',
+           'sample_proportions', 'minimize']
 
 import numpy as np
 import pandas as pd
@@ -105,6 +106,23 @@ def plot_normal_cdf(rbound=None, lbound=None, mean=0, sd=1):
 plot_cdf_area = plot_normal_cdf
 
 
+def sample_proportions(sample_size, probabilities):
+    """Return the proportion of random draws for each outcome in a distribution.
+
+    This function is similar to np.random.multinomial, but returns proportions
+    instead of counts.
+
+    Args:
+        ``sample_size``: The size of the sample to draw from the distribution.
+
+        ``probabilities``: An array of probabilities that forms a distribution.
+
+    Returns:
+        An array with the same length as ``probability`` that sums to 1.
+    """
+    return np.random.multinomial(sample_size, probabilities) / sample_size
+
+
 def proportions_from_distribution(table, label, sample_size,
                                   column_name='Random Sample'):
     """
@@ -115,8 +133,6 @@ def proportions_from_distribution(table, label, sample_size,
     from the distribution in ``table.column(label)``, then divides by
     ``sample_size`` to create the resulting column of proportions.
 
-    Returns a new ``Table`` and does not modify ``table``.
-
     Args:
         ``table``: An instance of ``Table``.
 
@@ -136,8 +152,7 @@ def proportions_from_distribution(table, label, sample_size,
         ``ValueError``: If the ``label`` is not in the table, or if
             ``table.column(label)`` does not sum to 1.
     """
-    proportions = (np.random.multinomial(sample_size, table.column(label)) /
-                   sample_size)
+    proportions = sample_proportions(sample_size, table.column(label))
     return table.with_column('Random Sample', proportions)
 
 
@@ -225,4 +240,3 @@ def objective(args):
         return result.x.item(0)
     else:
         return result.x
-

datascience/version.py

@@ -1 +1 @@
-__version__ = '0.10.3'
+__version__ = '0.10.4'

tests/test_util.py

@@ -66,3 +66,19 @@ def test_minimize_smooth():
 def test_minimize_array():
     assert _round_eq(2, ds.minimize(lambda x: (x[0]-2)**2, [0], array=True))
     assert _round_eq([2, 1], list(ds.minimize(lambda x: (x[0]-2)**2 + (x[1]-1)**2, [0, 0], array=True)))
+
+
+def test_sample_proportions():
+    uniform = ds.sample_proportions(1000, np.ones(50)/50)
+    assert len(uniform) == 50 and _round_eq(1, sum(uniform))
+    assert [x in (0, 0.5, 1) for x in ds.sample_proportions(2, ds.make_array(.2, .3, .5))]
+
+
+def test_proportions_from_distribution():
+    t = ds.Table().with_column('probs', np.ones(50)/50)
+    u = ds.proportions_from_distribution(t, 'probs', 1000)
+    assert t.num_columns == 1 and t.num_rows == 50
+    assert u.num_columns == 2 and u.num_rows == 50
+    uniform = u.column(1)
+    assert len(uniform) == 50 and _round_eq(1, sum(uniform))
+    assert [x in (0, 0.5, 1) for x in ds.sample_proportions(2, ds.make_array(.2, .3, .5))]