Fix random answers in DCM tests

This reverts commit 9c49885.
UDST · Aug 11, 2015 · 25c6b99 · 25c6b99
1 parent e22f053
commit 25c6b99
Showing 1 changed file with 22 additions and 24 deletions.
diff --git a/urbansim/models/tests/test_dcm.py b/urbansim/models/tests/test_dcm.py
@@ -163,7 +163,7 @@ def test_mnl_dcm(seed, basic_dcm, choosers, alternatives):
     pdt.assert_series_equal(
         choices,
         pd.Series(
-            ['b', 'j', 'g'], index=pd.Index([1, 3, 4], name='chooser_id')))
+            ['h', 'c', 'f'], index=pd.Index([1, 3, 4], name='chooser_id')))
 
     # check that we can do a YAML round-trip
     yaml_str = basic_dcm.to_yaml()
@@ -245,7 +245,6 @@ def test_mnl_dcm_yaml(basic_dcm, choosers, alternatives):
 
 def test_mnl_dcm_prob_mode_single(seed, basic_dcm_fit, choosers, alternatives):
     basic_dcm_fit.probability_mode = 'single_chooser'
-    basic_dcm_fit.choice_mode = 'aggregate'
 
     filtered_choosers, filtered_alts = basic_dcm_fit.apply_predict_filters(
         choosers, alternatives)
@@ -255,15 +254,15 @@ def test_mnl_dcm_prob_mode_single(seed, basic_dcm_fit, choosers, alternatives):
     pdt.assert_series_equal(
         probs,
         pd.Series(
-            [0.33709756,
-             0.23459563,
-             0.16326167,
-             0.11361837,
-             0.05502717,
-             0.03829495,
-             0.02665053,
-             0.01854685,
-             0.01290727],
+            [0.25666709612190147,
+             0.20225620916965448,
+             0.15937989234214262,
+             0.1255929308043417,
+             0.077988133629030815,
+             0.061455420294827229,
+             0.04842747874412457,
+             0.038161332007195688,
+             0.030071506886781514],
             index=pd.MultiIndex.from_product(
                 [[1], filtered_alts.index.values],
                 names=['chooser_id', 'alternative_id'])))
@@ -277,7 +276,6 @@ def test_mnl_dcm_prob_mode_single(seed, basic_dcm_fit, choosers, alternatives):
 def test_mnl_dcm_prob_mode_single_prediction_sample_size(
         seed, basic_dcm_fit, choosers, alternatives):
     basic_dcm_fit.probability_mode = 'single_chooser'
-    basic_dcm_fit.choice_mode = 'aggregate'
     basic_dcm_fit.prediction_sample_size = 5
 
     filtered_choosers, filtered_alts = basic_dcm_fit.apply_predict_filters(
@@ -288,19 +286,19 @@ def test_mnl_dcm_prob_mode_single_prediction_sample_size(
     pdt.assert_series_equal(
         probs,
         pd.Series(
-            [0.06805838,
-             0.20192435,
-             0.022939,
-             0.29015121,
-             0.41692705],
+            [0.11137766,
+             0.05449957,
+             0.14134044,
+             0.22761617,
+             0.46516616],
             index=pd.MultiIndex.from_product(
-                [[1], ['f', 'c', 'i', 'b', 'a']],
+                [[1], ['g', 'j', 'f', 'd', 'a']],
                 names=['chooser_id', 'alternative_id'])))
 
     sprobs = basic_dcm_fit.summed_probabilities(choosers, alternatives)
     pdt.assert_index_equal(
         sprobs.index,
-        pd.Index(['j', 'b', 'g', 'h', 'd'], name='alternative_id'))
+        pd.Index(['d', 'g', 'a', 'c', 'd'], name='alternative_id'))
     npt.assert_allclose(sprobs.sum(), len(filtered_choosers))
 
 
@@ -333,7 +331,7 @@ def test_mnl_dcm_choice_mode_agg(seed, basic_dcm_fit, choosers, alternatives):
 
     pdt.assert_series_equal(
         choices,
-        pd.Series(['a', 'c', 'b', 'd'], index=[0, 1, 3, 4]))
+        pd.Series(['f', 'a', 'd', 'c'], index=[0, 1, 3, 4]))
 
 
 def test_mnl_dcm_group(seed, grouped_choosers, alternatives):
@@ -382,7 +380,7 @@ def test_mnl_dcm_group(seed, grouped_choosers, alternatives):
     pdt.assert_series_equal(
         choices,
         pd.Series(
-            ['c', 'e', 'c', 'i'],
+            ['c', 'a', 'a', 'g'],
             index=pd.Index([0, 3, 1, 4], name='chooser_id')))
 
     # check that we don't get the same alt twice if they are removed
@@ -394,7 +392,7 @@ def test_mnl_dcm_group(seed, grouped_choosers, alternatives):
     pdt.assert_series_equal(
         choices,
         pd.Series(
-            ['c', 'e', 'd', 'i'],
+            ['c', 'a', 'b', 'g'],
             index=pd.Index([0, 3, 1, 4], name='chooser_id')))
 
 
@@ -462,7 +460,7 @@ def test_mnl_dcm_segmented(seed, grouped_choosers, alternatives):
     pdt.assert_series_equal(
         choices,
         pd.Series(
-            ['c', 'b', 'i', 'b', 'j'],
+            ['c', 'a', 'b', 'a', 'j'],
             index=pd.Index([0, 2, 3, 1, 4], name='chooser_id')))
 
     # check that we don't get the same alt twice if they are removed
@@ -474,7 +472,7 @@ def test_mnl_dcm_segmented(seed, grouped_choosers, alternatives):
     pdt.assert_series_equal(
         choices,
         pd.Series(
-            ['c', 'b', 'i', 'a', 'j'],
+            ['c', 'a', 'b', 'd', 'j'],
             index=pd.Index([0, 2, 3, 1, 4], name='chooser_id')))