Non-binary incidence (#127)

* update package version number as well * Allow non-binary incidence (#123) * Allow non-binary incidence * style * update tests to pass * add some progress indication * tidy up validation script, use histogram for a histogram * fix render and some typos Co-authored-by: bstabler <[email protected]>
ActivitySim · Oct 20, 2020 · 2c2536e · 2c2536e
1 parent 5c3a163
commit 2c2536e
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 25 deletions.
diff --git a/populationsim/balancer.py b/populationsim/balancer.py
@@ -208,7 +208,7 @@ def np_balancer(
                     yy + relaxed_constraint / float(importance))
 
             # update HH weights
-            weights_final[incidence[c] > 0] *= gamma[c]
+            weights_final *= pow(gamma[c], incidence[c])
 
             # clip weights to upper and lower bounds
             weights_final = np.clip(weights_final, weights_lower_bound, weights_upper_bound)

diff --git a/populationsim/simul_balancer.py b/populationsim/simul_balancer.py
@@ -238,7 +238,7 @@ def np_simul_balancer(
                         yy + (relaxed_constraint / float(importance)))
 
                 # update HH weights
-                sub_weights[z][incidence[c] > 0] *= gamma[z, c]
+                sub_weights[z] *= pow(gamma[z, c], incidence[c])
 
                 # clip weights to upper and lower bounds
                 sub_weights[z] = np.clip(sub_weights[z], weights_lower_bound, weights_upper_bound)

diff --git a/populationsim/steps/sub_balancing.py b/populationsim/steps/sub_balancing.py
@@ -252,9 +252,12 @@ def sub_balancing(settings, crosswalk, control_spec, incidence_table):
         # only want ones for which there are (non-zero) controls
         parent_ids = parent_controls_df.index.intersection(parent_ids)
 
-        for parent_id in parent_ids:
+        num_parent_ids = len(parent_ids)
+        for idx, parent_id in enumerate(parent_ids, start=1):
 
-            logger.info("balancing seed %s, %s %s" % (seed_id, parent_geography, parent_id))
+            log_msg = "balancing {}/{} seed {}, {} {}"
+            log_msg = log_msg.format(idx, num_parent_ids, seed_id, parent_geography, parent_id)
+            logger.info(log_msg)
 
             initial_weights = weights_df[weights_df[parent_geography] == parent_id]
             initial_weights = initial_weights.set_index(settings.get('household_id_col'))

diff --git a/populationsim/tests/test_steps.py b/populationsim/tests/test_steps.py
@@ -38,7 +38,7 @@ def teardown_function(func):
 
 
 TAZ_COUNT = 36
-TAZ_100_HH_COUNT = 25
+TAZ_100_HH_COUNT = 33
 TAZ_100_HH_REPOP_COUNT = 26
 
 
@@ -51,7 +51,7 @@ def test_full_run1():
         'meta_control_factoring',
         'final_seed_balancing',
         'integerize_final_seed_weights',
-        'sub_balancing.geography = TRACT',
+        'sub_balancing.geography=TRACT',
         'sub_balancing.geography=TAZ',
         'expand_households',
         'summarize',

diff --git a/scripts/validation.ipynb b/scripts/validation.ipynb