Enable more capability in neighborhood explorer.

paper/paper.md

@@ -181,11 +181,11 @@ Prior work suggests that historic trends would anticipate continued increases in
 Prior work expects that larger insured units will reduce risk [@knight_developing_2010] and we similarly observe that the claims rate decreases as the acrage included in an insured unit grows. However, after attempting multiple insured unit sizes including removal of smaller Optional Units [@zulauf_importance_2023] in post-hoc simulations, a gap persists in claims rates between the counterfactual and expected climate change simulations, suggesting our concerns may remain relevant across different risk units.
 
 ## Geographic bias
-We find significant results ($p < 0.05 / n$) more common on the eastern side of the study area as shown in Figure @fig:geo.
+We find significant results ($p < 0.05 / n$) more common in some areas as shown in Figure @fig:geo.
 
 ![Example simulation in our interactive tool’s geographic view. Our projections vary across different geographic areas.](./img/map.png "Example simulation in our interactive tool’s geographic view. Our projections vary across different geographic areas."){ width=85% #fig:geo }
 
-This spatial bias may partially reflect that a number of more western neighborhoods have less land dedicated to maize so simulate with smaller sample sizes and fail to reach significance. In other cases, this geographic effect may also reflect disproportionate stress or other changes relative to historic conditions. In particular, as further explorable in our interactive tools, we note some east / west bias in changes to precipitation, temperature, and VPD / SVP.
+This spatial bias may partially reflect that a number of neighborhoods have less land dedicated to maize so simulate with smaller sample sizes and fail to reach significance. In other cases, this geographic effect may also reflect disproportionate stress or other changes relative to historic conditions. In particular, as further explorable in our interactive tools, we note some geographic bias in changes to precipitation, temperature, and VPD / SVP.
 
 ## Heat and drought stress
 Our model shows depressed yields during combined warmer and drier conditions potentially similar to 2012 which saw poor US maize outcomes [@ers_weather_2013].

paper/viz/buttons.py

@@ -242,27 +242,25 @@ def __init__(self, sketch, x, y, initial_config, on_change):
             0,
             current_y,
             'Scenario',
-            ['2030 conditions', '2050 conditions'],
+            ['2030 series', '2050 series'],
             self._config.get_scenario(),
             lambda x: self._set_config(self._config.get_with_scenario(x))
         )
 
-#        current_y += self._scenario_buttons.get_height() + 12
-#
-#        self._range_buttons = ToggleButtonSet(
-#            self._sketch,
-#            0,
-#            current_y,
-#            'Range of Risk',
-#            ['1 year', '3 years'],
-#            self._config.get_risk_range(),
-#            lambda x: self._set_config(self._config.get_with_risk_range(x))
-#        )
-#
-#        current_y += self._range_buttons.get_height() + 30
-
         current_y += self._scenario_buttons.get_height() + 12
 
+        self._range_buttons = ToggleButtonSet(
+            self._sketch,
+            0,
+            current_y,
+            'Range of Risk',
+            ['Sample 1 Year', 'Avg All Years'],
+            self._config.get_risk_range(),
+            lambda x: self._set_config(self._config.get_with_risk_range(x))
+        )
+
+        current_y += self._range_buttons.get_height() + 30
+
         self._threshold_buttons = ToggleButtonSet(
             self._sketch,
             0,
@@ -356,7 +354,7 @@ def step(self, mouse_x, mouse_y, clicked):
         mouse_y = mouse_y - self._y
 
         self._scenario_buttons.step(mouse_x, mouse_y, clicked)
-#        self._range_buttons.step(mouse_x, mouse_y, clicked)
+        self._range_buttons.step(mouse_x, mouse_y, clicked)
         self._metric_buttons.step(mouse_x, mouse_y, clicked)
         self._viz_buttons.step(mouse_x, mouse_y, clicked)
         self._threshold_buttons.step(mouse_x, mouse_y, clicked)

paper/viz/const.py

@@ -33,8 +33,8 @@
 YIELD_CHANGE_MAX_VALUE = 0.4
 YIELD_CHANGE_INCREMENT = 0.05
 
-RISK_MIN_VALUE = -30
-RISK_MAX_VALUE = 60
+RISK_MIN_VALUE = -20
+RISK_MAX_VALUE = 50
 RISK_INCREMENT = 10
 
 STD_MIN_VALUE = -30

paper/viz/data_struct.py

@@ -54,23 +54,18 @@ def combine(self, other):
         elif other_count == 0:
             return self
 
-        self_weighted_p = self.get_p_value() * self_count
-        other_weighted_p = other.get_p_value() * other_count
-        new_p = (self_weighted_p + other_weighted_p) / (self_count + other_count)
+        new_p = min([self.get_p_value(), other.get_p_value()])
 
-        def combine_probs(a, b):
-            a_not = 1 - a
-            b_not = 1 - b
-            not_a_b = a_not * b_not
-            return 1 - not_a_b
+        def combine_probs(a, a_count, b, b_count):
+            return (a * a_count + b * b_count) / (a_count + b_count)
 
         self_control_risk = self.get_control_risk()
         other_control_risk = other.get_control_risk()
-        new_control_risk = combine_probs(self_control_risk, other_control_risk)
+        new_control_risk = combine_probs(self_control_risk, self_count, other_control_risk, other_count)
 
         self_experimental_risk = self.get_experimental_risk()
         other_experimental_risk = other.get_experimental_risk()
-        new_experimental_risk = combine_probs(self_experimental_risk, other_experimental_risk)
+        new_experimental_risk = combine_probs(self_experimental_risk, self_count, other_experimental_risk, other_count)
 
         new_count = self_count + other_count
 
@@ -152,13 +147,11 @@ def combine(self, other):
         elif other_count == 0:
             return self
 
-        self_weighted_p = self.get_p_value() * self_count
-        other_weighted_p = other.get_p_value() * other_count
-        weighted_p = (self_weighted_p + other_weighted_p) / (self_count + other_count)
+        new_p = min([self.get_p_value(), other.get_p_value()])
         return YieldComparison(
             self.get_prior().combine(other.get_prior()),
             self.get_predicted().combine(other.get_predicted()),
-            weighted_p
+            new_p
         )
 
 

paper/viz/preprocess.py

@@ -107,14 +107,13 @@ def get_scenario_year(record):
 
 
 def make_scatter_values(records, climate_deltas, configuration):
-    count = len(records)
     scenario = configuration.get_scenario()
     target_year = int(scenario[:4])
     target_loss = '25% loss' if configuration.get_loss() == '75% cov' else '15% loss'
     scenario_records = filter(lambda x: scenario.startswith(str(get_scenario_year(x))), records)
     loss_records = filter(lambda x: x.get_loss() == target_loss, scenario_records)
 
-    if configuration.get_risk_range() == '1 year':
+    if configuration.get_risk_range() == 'Sample 1 Year':
         year_records = filter(lambda x: x.get_year() in [2030, 2050], loss_records)
     else:
         year_records_nested = toolz.itertoolz.reduceby(
@@ -124,6 +123,9 @@ def make_scatter_values(records, climate_deltas, configuration):
         )
         year_records = year_records_nested.values()
 
+    year_records = list(year_records)
+    count = len(year_records)
+
     p_threshold_naive = {
         'p < 0.05': 0.05,
         'p < 0.10': 0.1

paper/viz/render_images.sh

@@ -1,4 +1,4 @@
 python3 hist_viz.py ../outputs/export_hist.csv 2050 75 ../img/hist.png
-python3 results_viz_entry.py ../outputs/export_summary.csv 2050 risk map 0.05 Bonferroni significant none jul 75 ../outputs/export_climate.csv ../img/map.png
-python3 results_viz_entry.py ../outputs/export_summary.csv 2050 risk scatter 0.05 Bonferroni significant chirps jul 75 ../outputs/export_climate.csv ../img/scatter.png
+python3 results_viz_entry.py ../outputs/export_summary.csv 2050 risk map 0.05 Bonferroni all none jul 75 ../outputs/export_climate.csv ../img/map.png
+python3 results_viz_entry.py ../outputs/export_summary.csv 2030 risk scatter 0.05 Bonferroni significant chirps jul 75 ../outputs/export_climate.csv ../img/scatter.png
 python3 std_hist.py ../outputs/export_combined_tasks.csv ../img/std.png

paper/viz/results_viz.py

@@ -47,8 +47,8 @@ def __init__(self, target, loading_id, default_configuration=None, data_loc=None
             self._config = default_configuration
         else:
             self._config = buttons.Configuration(
-                '2050 conditions',
-                '1 year',
+                '2050 series',
+                'Avg All Years',
                 'yield',
                 'scatter',
                 'p < 0.05',
@@ -243,16 +243,10 @@ def _draw_annotation(self):
     def _get_description(self):
         metric = self._config.get_metric()
 
-        if self._config.get_risk_range() == '3 years':
-            if metric == 'yield':
-                agg_str = 'Averaging across 3 years.'
-            else:
-                agg_str = 'Prob of event in any of 3 years.'
+        if self._config.get_risk_range() == '':
+            agg_str = 'Averaging across all years.'
         else:
-            if metric == 'yield':
-                agg_str = 'Single year avg yields.'
-            else:
-                agg_str = 'Prob of event in single year.'
+            agg_str = 'Single year avg yields.'
 
         in_map = self._config.get_visualization() == 'map'
         no_var = self._config.get_var() == 'no var'

paper/viz/results_viz_entry.py

@@ -29,7 +29,7 @@ def main():
         sys.exit(1)
     else:
         data_loc = sys.argv[1]
-        scenario = sys.argv[2] + ' conditions'
+        scenario = sys.argv[2] + ' series'
         metric = sys.argv[3]
         visualization = sys.argv[4]
         threshold = 'p < ' + sys.argv[5]
@@ -43,7 +43,7 @@ def main():
 
         default_configuration = buttons.Configuration(
             scenario,
-            '1 year',
+            'Avg All Years',
             metric,
             visualization,
             threshold,