Add Likelihood of stop

docs/data-import.rst

@@ -25,8 +25,8 @@ To load an existing database dump on S3, run:
 .. code-block:: bash
 
     dropdb traffic_stops_nc && createdb -E UTF-8 traffic_stops_nc
-    aws s3 cp s3://forwardjustice-trafficstops/trafficstops-staging_database.dump .
-    pg_restore -Ox -d traffic_stops_nc trafficstops-staging_database.dump
+    aws s3 cp s3://forwardjustice-trafficstops/traffic_stops_nc-20250204.dump .
+    pg_restore -Ox -d traffic_stops_nc traffic_stops_nc-20250204.dump
 
 
 Raw NC Data (slower)

nc/admin.py

@@ -2,7 +2,15 @@
 from django import forms
 from django.contrib import admin
 
-from nc.models import Agency, Resource, ResourceFile, StopSummary
+from nc.models import (
+    Agency,
+    ContrabandSummary,
+    LikelihoodStopSummary,
+    NCCensusProfile,
+    Resource,
+    ResourceFile,
+    StopSummary,
+)
 
 
 class AgencyAdmin(admin.ModelAdmin):
@@ -45,6 +53,24 @@ def agency_name(self, obj):
         return obj.agency.name
 
 
+@admin.register(ContrabandSummary)
+class ContrabandSummaryAdmin(admin.ModelAdmin):
+    date_hierarchy = "date"
+    list_display = (
+        "id",
+        "date",
+        "agency",
+        "stop_purpose_group",
+        "driver_searched",
+        "driver_arrest",
+        "contraband_found",
+    )
+    list_filter = ("date", "stop_purpose_group")
+    list_select_related = ("agency",)
+    search_fields = ("id", "agency__name")
+    raw_id_fields = ("stop", "agency", "search", "contraband")
+
+
 class InlineResourceFile(admin.StackedInline):
     model = ResourceFile
     fields = (
@@ -95,6 +121,74 @@ def save_related(self, request, form, formsets, change):
         form.instance.agencies.set(form.cleaned_data["agencies"], clear=True)
 
 
+@admin.register(NCCensusProfile)
+class NCCensusProfileAdmin(admin.ModelAdmin):
+    list_display = (
+        "id",
+        "acs_id",
+        "location",
+        "geography",
+        "race",
+        "population",
+        "population_total",
+        "population_pct",
+    )
+    list_filter = ("geography", "race")
+    ordering = ("location",)
+    readonly_fields = (
+        "acs_id",
+        "location",
+        "geography",
+        "race",
+        "population",
+        "population_total",
+        "population_percent",
+        "source",
+    )
+    search_fields = ("location", "id", "acs_id")
+
+    @admin.display(ordering="population_percent")
+    def population_pct(self, obj):
+        return f"{obj.population_percent:.2%}"
+
+
+@admin.register(LikelihoodStopSummary)
+class LikelihoodStopSummaryAdmin(admin.ModelAdmin):
+    list_display = (
+        "id",
+        "agency_name",
+        "driver_race_comb",
+        "population",
+        "population_total",
+        "stops",
+        "stops_total",
+        "stop_rate",
+        "baseline_rate",
+        "stop_rate_ratio",
+    )
+    list_filter = ("driver_race_comb", "agency")
+    list_select_related = ("agency",)
+    readonly_fields = (
+        "id",
+        "agency",
+        "driver_race_comb",
+        "population",
+        "population_total",
+        "population_percent",
+        "stops",
+        "stops_total",
+        "stop_rate",
+        "baseline_rate",
+        "stop_rate_ratio",
+    )
+    search_fields = ("agency__name",)
+    ordering = ("agency__name", "driver_race_comb")
+
+    @admin.display(ordering="agency__name")
+    def agency_name(self, obj):
+        return obj.agency.name
+
+
 admin.site.register(Agency, AgencyAdmin)
 admin.site.register(StopSummary, StopSummaryAdmin)
 admin.site.register(Resource, ResourceAdmin)

nc/apps.py

@@ -0,0 +1,7 @@
+from django.apps import AppConfig
+
+
+class InventoryConfig(AppConfig):
+    default_auto_field = "django.db.models.AutoField"
+    name = "nc"
+    verbose_name = "North Carolina"

nc/migrations/0013_nccensusprofile.py

@@ -0,0 +1,31 @@
+# Generated by Django 3.2.25 on 2025-02-03 16:44
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('nc', '0012_contrabandsummary'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='NCCensusProfile',
+            fields=[
+                ('id', models.AutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('acs_id', models.CharField(max_length=32, verbose_name='ACS ID')),
+                ('location', models.CharField(max_length=64)),
+                ('geography', models.CharField(choices=[('state', 'State'), ('county', 'County'), ('place', 'Place')], max_length=16)),
+                ('source', models.CharField(max_length=64)),
+                ('race', models.CharField(max_length=32)),
+                ('population', models.BigIntegerField()),
+                ('population_total', models.BigIntegerField()),
+                ('population_percent', models.FloatField()),
+            ],
+            options={
+                'verbose_name': 'NC Census Profile',
+                'verbose_name_plural': 'NC Census Profiles',
+            },
+        ),
+    ]

nc/migrations/0014_auto_20250204_1040.py

@@ -0,0 +1,53 @@
+# Generated by Django 3.2.25 on 2025-02-04 15:40
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("nc", "0013_nccensusprofile"),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="LikelihoodStopSummary",
+            fields=[
+                (
+                    "id",
+                    models.PositiveIntegerField(
+                        primary_key=True, serialize=False, verbose_name="ID"
+                    ),
+                ),
+                (
+                    "driver_race_comb",
+                    models.CharField(
+                        choices=[
+                            ("Asian", "Asian"),
+                            ("Black", "Black"),
+                            ("Hispanic", "Hispanic"),
+                            ("Native American", "Native American"),
+                            ("Other", "Other"),
+                            ("White", "White"),
+                        ],
+                        db_column="driver_race",
+                        max_length=216,
+                        verbose_name="Driver Race",
+                    ),
+                ),
+                ("population", models.PositiveIntegerField()),
+                ("population_total", models.PositiveIntegerField()),
+                ("population_percent", models.DecimalField(decimal_places=2, max_digits=5)),
+                ("stops", models.PositiveIntegerField()),
+                ("stops_total", models.PositiveIntegerField()),
+                ("stop_rate", models.DecimalField(decimal_places=2, max_digits=5)),
+                ("baseline_rate", models.DecimalField(decimal_places=2, max_digits=5)),
+                ("stop_rate_ratio", models.DecimalField(decimal_places=2, max_digits=5)),
+            ],
+            options={
+                "verbose_name": "Likelihood of Stop Summary",
+                "verbose_name_plural": "Likelihood of Stop Summaries",
+                "managed": False,
+            },
+        ),
+    ]

nc/models.py

@@ -442,3 +442,105 @@ def __str__(self):
         if self.file:
             return f"{self.file.name} for {self.resource.title}"
         return f"Resource file for {self.resource.title}"
+
+
+class NCCensusProfile(models.Model):
+    class GeographyChoices(models.TextChoices):
+        STATE = "state", "State"
+        COUNTY = "county", "County"
+        PLACE = "place", "Place"
+
+    acs_id = models.CharField(verbose_name="ACS ID", max_length=32)
+    location = models.CharField(max_length=64)
+    geography = models.CharField(max_length=16, choices=GeographyChoices.choices)
+    source = models.CharField(max_length=64)
+    race = models.CharField(max_length=32)
+    population = models.BigIntegerField()
+    population_total = models.BigIntegerField()
+    population_percent = models.FloatField()
+
+    class Meta:
+        verbose_name = "NC Census Profile"
+        verbose_name_plural = "NC Census Profiles"
+
+    def __str__(self):
+        return f"{self.location} {self.race} people ({self.geography})"
+
+
+class Race(models.TextChoices):
+    ASIAN = "Asian"
+    BLACK = "Black"
+    HISPANIC = "Hispanic"
+    NATIVE_AMERICAN = "Native American"
+    OTHER = "Other"
+    WHITE = "White"
+
+
+LIKELIHOOD_STOP_SQL = """
+WITH stop_summary AS (
+    SELECT
+        agency_id
+        , name AS agency
+        , agency.census_profile_id AS acs_id
+        , driver_race_comb AS driver_race
+        , sum(count) AS stops
+        , sum(sum(count)) OVER (PARTITION BY agency_id)::integer AS stops_total
+        , (sum(count) * 1.0) / sum(sum(count)) OVER (PARTITION BY agency_id) AS stops_percent
+    FROM nc_stopsummary summary
+    JOIN nc_agency agency ON (summary.agency_id = agency.id)
+    WHERE agency.census_profile_id IS NOT NULL
+    GROUP BY 1, 2, 3, 4
+    ORDER BY 2, 3
+), stop_summary_with_acs AS (
+    SELECT
+        stops.*
+        , acs.*
+        , (stops * 1.0) / NULLIF(population, 0) AS stop_rate
+    FROM stop_summary stops
+    JOIN nc_nccensusprofile acs ON (stops.acs_id = acs.acs_id AND stops.driver_race = acs.race)
+    ORDER BY agency_id, driver_race
+)
+SELECT
+    row_number() over() AS id
+    , parent.agency_id
+    , parent.agency
+    , parent.driver_race
+    , parent.population
+    , parent.population_total
+    , parent.population_percent
+    , parent.stops
+    , parent.stops_total
+    , parent.stop_rate
+    , baseline.stop_rate AS baseline_rate
+    , (parent.stop_rate - baseline.stop_rate) / baseline.stop_rate AS stop_rate_ratio
+FROM stop_summary_with_acs parent
+JOIN stop_summary_with_acs baseline ON (baseline.driver_race = 'White' AND parent.agency = baseline.agency)
+ORDER BY agency_id, driver_race
+"""  # noqa
+
+
+class LikelihoodStopSummary(pg.ReadOnlyMaterializedView):
+    sql = LIKELIHOOD_STOP_SQL
+    # Don't create view with data, this will be manually managed
+    # and refreshed by the data import process
+    # https://github.com/mikicz/django-pgviews#with-no-data
+    with_data = False
+
+    id = models.PositiveIntegerField(verbose_name="ID", primary_key=True)
+    agency = models.ForeignKey("Agency", on_delete=models.DO_NOTHING)
+    driver_race_comb = models.CharField(
+        verbose_name="Driver Race", max_length=216, choices=Race.choices, db_column="driver_race"
+    )
+    population = models.PositiveIntegerField()
+    population_total = models.PositiveIntegerField()
+    population_percent = models.DecimalField(max_digits=5, decimal_places=2)
+    stops = models.PositiveIntegerField()
+    stops_total = models.PositiveIntegerField()
+    stop_rate = models.DecimalField(max_digits=5, decimal_places=2)
+    baseline_rate = models.DecimalField(max_digits=5, decimal_places=2)
+    stop_rate_ratio = models.DecimalField(max_digits=5, decimal_places=2)
+
+    class Meta:
+        managed = False
+        verbose_name = "Likelihood of Stop Summary"
+        verbose_name_plural = "Likelihood of Stop Summaries"

nc/urls.py

@@ -125,4 +125,9 @@
         views.AgencyArrestsPercentageOfStopsPerContrabandTypeView.as_view(),
         name="arrests-percentage-of-stops-per-contraband-type",
     ),
+    path(
+        "api/agency/<agency_id>/likelihood-of-stops/",
+        views.LikelihoodStopView.as_view(),
+        name="likelihood-of-stops",
+    ),
 ]

nc/views/__init__.py

@@ -1,2 +1,3 @@
 from .arrests import *  # noqa
+from .likelihood import *  # noqa
 from .main import *  # noqa