Merge branch 'develop' into fix_default_settings_current_date

Author: jlubken

tests/conftest.py

@@ -0,0 +1,114 @@
+from datetime import datetime
+
+import pytest
+import pandas as pd
+
+from src.penn_chime.parameters import (
+    Parameters,
+    Disposition,
+    Regions,
+)
+from src.penn_chime.models import SimSirModel
+
+
+class MockStreamlit:
+    """Mock implementation of streamlit
+
+    We just want to verify that st _attempted_ to render the right stuff
+    so we store the input, and make sure that it matches what we expect
+    """
+
+    def __init__(self):
+        self.render_store = []
+        self.markdown = self.just_store_instead_of_rendering
+        self.latex = self.just_store_instead_of_rendering
+        self.subheader = self.just_store_instead_of_rendering
+
+    def just_store_instead_of_rendering(self, inp, *args, **kwargs):
+        self.render_store.append(inp)
+        return None
+
+
+@pytest.fixture
+def mock_st():
+    return MockStreamlit()
+
+
+# The defaults in settings will change and break the tests
+@pytest.fixture
+def DEFAULTS():
+    return Parameters(
+        region=Regions(
+            delaware=564696,
+            chester=519293,
+            montgomery=826075,
+            bucks=628341,
+            philly=1581000,
+        ),
+        current_date=datetime(year=2020, month=3, day=28),
+        current_hospitalized=14,
+        date_first_hospitalized=datetime(year=2020, month=3, day=7),
+        doubling_time=4.0,
+        n_days=60,
+        market_share=0.15,
+        relative_contact_rate=0.3,
+        hospitalized=Disposition(0.025, 7),
+        icu=Disposition(0.0075, 9),
+        ventilated=Disposition(0.005, 10),
+    )
+
+
+@pytest.fixture
+def param():
+    return Parameters(
+        current_date=datetime(year=2020, month=3, day=28),
+        current_hospitalized=100,
+        doubling_time=6.0,
+        market_share=0.05,
+        relative_contact_rate=0.15,
+        population=500000,
+        hospitalized=Disposition(0.05, 7),
+        icu=Disposition(0.02, 9),
+        ventilated=Disposition(0.01, 10),
+        n_days=60,
+    )
+
+
+@pytest.fixture
+def halving_param():
+    return Parameters(
+        current_date=datetime(year=2020, month=3, day=28),
+        current_hospitalized=100,
+        doubling_time=6.0,
+        market_share=0.05,
+        relative_contact_rate=0.7,
+        population=500000,
+        hospitalized=Disposition(0.05, 7),
+        icu=Disposition(0.02, 9),
+        ventilated=Disposition(0.01, 10),
+        n_days=60,
+    )
+
+
+@pytest.fixture
+def model(param):
+    return SimSirModel(param)
+
+
+@pytest.fixture
+def halving_model(halving_param):
+    return SimSirModel(halving_param)
+
+
+@pytest.fixture
+def admits_df():
+    return pd.read_csv(
+        "tests/by_doubling_time/2020-03-28_projected_admits.csv", parse_dates=["date"]
+    )
+
+
+@pytest.fixture
+def census_df():
+    return pd.read_csv(
+        "tests/by_doubling_time/2020-03-28_projected_census.csv", parse_dates=["date"]
+    )

tests/penn_chime/test_charts.py

@@ -0,0 +1,64 @@
+from math import ceil
+from datetime import datetime
+
+import altair as alt
+import pytest
+
+from src.penn_chime.charts import (
+    build_admits_chart,
+    build_census_chart,
+    build_descriptions,
+)
+
+# TODO add test for asterisk
+
+
+def test_admits_chart(admits_df):
+    chart = build_admits_chart(alt=alt, admits_df=admits_df)
+    assert isinstance(chart, (alt.Chart, alt.LayerChart))
+    assert round(chart.data.iloc[40].icu, 0) == 39
+
+    # test fx call with no params
+    with pytest.raises(TypeError):
+        build_admits_chart()
+
+
+def test_build_descriptions(admits_df, param):
+    chart = build_admits_chart(alt=alt, admits_df=admits_df)
+    description = build_descriptions(chart=chart, labels=param.labels)
+
+    hosp, icu, vent = description.split("\n\n")  # break out the description into lines
+
+    max_hosp = chart.data["hospitalized"].max()
+    assert str(ceil(max_hosp)) in hosp
+
+
+def test_no_asterisk(admits_df, param):
+    param.n_days = 600
+
+    chart = build_admits_chart(alt=alt, admits_df=admits_df)
+    description = build_descriptions(chart=chart, labels=param.labels)
+    assert "*" not in description
+
+
+def test_census(census_df, param):
+    chart = build_census_chart(alt=alt, census_df=census_df)
+    description = build_descriptions(chart=chart, labels=param.labels)
+
+    assert str(ceil(chart.data["ventilated"].max())) in description
+    assert str(chart.data["icu"].idxmax()) not in description
+    assert (
+        datetime.strftime(chart.data.iloc[chart.data["icu"].idxmax()].date, "%b %d")
+        in description
+    )
+
+
+def test_census_chart(census_df):
+    chart = build_census_chart(alt=alt, census_df=census_df)
+    assert isinstance(chart, (alt.Chart, alt.LayerChart))
+    assert chart.data.iloc[1].hospitalized == 3
+    assert chart.data.iloc[49].ventilated == 365
+
+    # test fx call with no params
+    with pytest.raises(TypeError):
+        build_census_chart()

tests/penn_chime/test_models.py

@@ -0,0 +1,178 @@
+from datetime import date
+
+import pytest
+import pandas as pd
+import numpy as np
+
+from src.penn_chime.models import (
+    sir,
+    sim_sir_df,
+    get_growth_rate,
+)
+
+from src.penn_chime.constants import EPSILON
+
+
+def test_sir():
+    """
+    Someone who is good at testing, help
+    """
+    sir_test = sir(100, 1, 0, 0.2, 0.5, 1)
+    assert sir_test == (
+        0.7920792079207921,
+        0.20297029702970298,
+        0.0049504950495049506,
+    ), "This contrived example should work"
+
+    assert isinstance(sir_test, tuple)
+    for v in sir_test:
+        assert isinstance(v, float)
+        assert v >= 0
+
+    # Certain things should *not* work
+    with pytest.raises(TypeError) as error:
+        sir("S", 1, 0, 0.2, 0.5, 1)
+    assert str(error.value) == "can't multiply sequence by non-int of type 'float'"
+
+    with pytest.raises(TypeError) as error:
+        sir(100, "I", 0, 0.2, 0.5, 1)
+    assert str(error.value) == "can't multiply sequence by non-int of type 'float'"
+
+    with pytest.raises(TypeError) as error:
+        sir(100, 1, "R", 0.2, 0.5, 1)
+    assert str(error.value) == "unsupported operand type(s) for +: 'float' and 'str'"
+
+    with pytest.raises(TypeError) as error:
+        sir(100, 1, 0, "beta", 0.5, 1)
+    assert str(error.value) == "bad operand type for unary -: 'str'"
+
+    with pytest.raises(TypeError) as error:
+        sir(100, 1, 0, 0.2, "gamma", 1)
+    assert str(error.value) == "unsupported operand type(s) for -: 'float' and 'str'"
+
+    with pytest.raises(TypeError) as error:
+        sir(100, 1, 0, 0.2, 0.5, "N")
+    assert str(error.value) == "unsupported operand type(s) for /: 'str' and 'float'"
+
+    # Zeros across the board should fail
+    with pytest.raises(ZeroDivisionError):
+        sir(0, 0, 0, 0, 0, 0)
+
+
+def test_sim_sir():
+    """
+    Rounding to move fast past decimal place issues
+    """
+    raw_df = sim_sir_df(
+        5, 6, 7, 0.1, 0, 0.1, 40,  # s  # i  # r  # gamma  # i_day  # beta1  # n_days1
+    )
+
+    first = raw_df.iloc[0, :]
+    last = raw_df.iloc[-1, :]
+
+    assert round(first.susceptible, 0) == 5
+    assert round(first.infected, 2) == 6
+    assert round(first.recovered, 0) == 7
+    assert round(last.susceptible, 2) == 0
+    assert round(last.infected, 2) == 0.18
+    assert round(last.recovered, 2) == 17.82
+
+    assert isinstance(raw_df, pd.DataFrame)
+
+
+def test_growth_rate():
+    assert np.round(get_growth_rate(5) * 100.0, decimals=4) == 14.8698
+    assert np.round(get_growth_rate(0) * 100.0, decimals=4) == 0.0
+    assert np.round(get_growth_rate(-4) * 100.0, decimals=4) == -15.9104
+
+
+def test_model(model, param):
+    # test the Model
+
+    assert round(model.infected, 0) == 45810.0
+    assert isinstance(model.infected, float)  # based off note in models.py
+
+    # test the class-calculated attributes
+    # we're talking about getting rid of detection probability
+    # assert model.detection_probability == 0.125
+    assert model.intrinsic_growth_rate == 0.12246204830937302
+    assert abs(model.beta - 4.21501347256401e-07) < EPSILON
+    assert model.r_t == 2.307298374881539
+    assert model.r_naught == 2.7144686763312222
+    assert model.doubling_time_t == 7.764405988534983
+
+
+def test_model_raw_start(model, param):
+    raw_df = model.raw_df
+
+    # test the things n_days creates, which in turn tests sim_sir, sir, and get_dispositions
+
+    # print('n_days: %s; i_day: %s' % (param.n_days, model.i_day))
+    assert len(raw_df) == (len(np.arange(-model.i_day, param.n_days + 1))) == 104
+
+    first = raw_df.iloc[0, :]
+    second = raw_df.iloc[1, :]
+
+    assert first.susceptible == 499600.0
+    assert round(second.infected, 0) == 449.0
+    assert list(model.dispositions_df.iloc[0, :]) == [
+        -43,
+        date(year=2020, month=2, day=14),
+        1.0,
+        0.4,
+        0.2,
+    ]
+    assert round(raw_df.recovered[30], 0) == 7083.0
+
+    d, dt, s, i, r = list(model.dispositions_df.iloc[60, :])
+    assert dt == date(year=2020, month=4, day=14)
+    assert [round(v, 0) for v in (d, s, i, r)] == [17, 549.0, 220.0, 110.0]
+
+
+def test_model_conservation(param, model):
+    raw_df = model.raw_df
+
+    assert (0.0 <= raw_df.susceptible).all()
+    assert (0.0 <= raw_df.infected).all()
+    assert (0.0 <= raw_df.recovered).all()
+
+    diff = raw_df.susceptible + raw_df.infected + raw_df.recovered - param.population
+    assert (diff < 0.1).all()
+
+    assert (raw_df.susceptible <= param.population).all()
+    assert (raw_df.infected <= param.population).all()
+    assert (raw_df.recovered <= param.population).all()
+
+
+def test_model_raw_end(param, model):
+    raw_df = model.raw_df
+    last = raw_df.iloc[-1, :]
+    assert round(last.susceptible, 0) == 83391.0
+
+
+def test_model_monotonicity(param, model):
+    raw_df = model.raw_df
+
+    # Susceptible population should be non-increasing, and Recovered non-decreasing
+    assert (raw_df.susceptible[1:] - raw_df.susceptible.shift(1)[1:] <= 0).all()
+    assert (raw_df.recovered[1:] - raw_df.recovered.shift(1)[1:] >= 0).all()
+
+
+def test_model_cumulative_census(param, model):
+    # test that census is being properly calculated
+    raw_df = model.raw_df
+    admits_df = model.admits_df
+    df = pd.DataFrame(
+        {
+            "hospitalized": admits_df.hospitalized,
+            "icu": admits_df.icu,
+            "ventilated": admits_df.ventilated,
+        }
+    )
+    admits = df.cumsum()
+
+    # TODO: is 1.0 for ceil function?
+    diff = admits.hospitalized[1:-1] - (
+        0.05 * 0.05 * (raw_df.infected[1:-1] + raw_df.recovered[1:-1]) - 1.0
+    )
+    assert (diff.abs() < 0.1).all()