Adding testing and validation

README.md

@@ -102,7 +102,7 @@ Note that the user-supplied configuration file is used to provide
 
 ### Usage examples:
 - Using the default emodl template: `python runScenarios.py
-  --running_location Local --region IL  --experiment_config sample_experiment.yaml`
+  --running_location Local --region IL  --experiment_config ./experiment_configs/sample_experiment.yaml`
 - Using a different emodl template: `python runScenarios.py
   --running_location Local  --region IL  --experiment_config sample_experiment.yaml --emodl_template simplemodel_testing.emodl`
 - Specifying experiment name suffix and changing running_location : `python runScenarios.py

dev-requirements.txt

@@ -0,0 +1 @@
+pytest>=5

extendedcobey.yaml → experiment_configs/extendedcobey.yaml

@@ -103,16 +103,16 @@ sampled_parameters:
     function_kwargs: {'low':0.95, 'high':1}
   'socialDistance_time1':
     custom_function: DateToTimestep
-    function_kwargs: {'dates': 2020-03-12}
+    function_kwargs: {'dates': 2020-03-12, 'startdate_col': 'startdate'}
   'socialDistance_time2':
     custom_function: DateToTimestep
-    function_kwargs: {'dates': 2020-03-17}
+    function_kwargs: {'dates': 2020-03-17, 'startdate_col': 'startdate'}
   'socialDistance_time3':
     custom_function: DateToTimestep
-    function_kwargs: {'dates': 2020-03-21}
+    function_kwargs: {'dates': 2020-03-21, 'startdate_col': 'startdate'}
   'socialDistanceSTOP_time':
       custom_function: DateToTimestep
-      function_kwargs: {'dates': 2020-04-24}
+      function_kwargs: {'dates': 2020-04-24, 'startdate_col': 'startdate'}
 fitted_parameters:
   # Note IL, NMH, Chicago and EMS_3 were fitted, the other EMS area Kis need updating/checking
   Kis:

runScenarios.py

@@ -20,6 +20,7 @@
 log = logging.getLogger(__name__)
 
 mpl.rcParams['pdf.fonttype'] = 42
+
 today = datetime.today()
 DEFAULT_CONFIG = './extendedcobey.yaml'
 
@@ -34,8 +35,9 @@ def _parse_config_parameter(df, parameter, parameter_function):
         function_name = parameter_function['custom_function']
         function_kwargs = parameter_function['function_kwargs']
         if function_name == 'DateToTimestep':
-            function_kwargs['startdate'] = first_day
-            return [DateToTimestep(**function_kwargs) for i in range(len(df))]
+            startdate_col = function_kwargs['startdate_col']
+            return [DateToTimestep(function_kwargs['dates'], df[startdate_col][i])
+                    for i in range(len(df))]
         elif function_name == 'subtract':
             return df[function_kwargs['x1']] - df[function_kwargs['x2']]
         else:
@@ -76,7 +78,7 @@ def add_config_parameter_column(df, parameter, parameter_function, age_bins=None
     df: pd.DataFrame
         dataframe with the additional column(s) added
     """
-    if parameter_function.get('expand_by_age'):
+    if isinstance(parameter_function, dict) and parameter_function.get('expand_by_age'):
         if not age_bins:
             raise ValueError("Ages bins must be specified if using an age expansion")
         if 'list' in parameter_function:
@@ -97,7 +99,7 @@ def add_config_parameter_column(df, parameter, parameter_function, age_bins=None
         else:
             raise ValueError(f"Unknown type of parameter {parameter} for expand_by_age")
     else:
-        if 'matrix' in parameter_function:
+        if isinstance(parameter_function, dict) and 'matrix' in parameter_function:
             m = parameter_function['matrix']
             for i, row in enumerate(m):
                 for j, item in enumerate(row):
@@ -138,6 +140,7 @@ def generateParameterSamples(samples, pop, first_day, config, age_bins):
     df['sample_num'] = range(samples)
     df['speciesS'] = pop
     df['initialAs'] = config['experiment_setup_parameters']['initialAs']
+    df['startdate'] = experiment_config['fixed_parameters_region_specific']['startdate'][region]
 
     for parameter, parameter_function in config['sampled_parameters'].items():
         df = add_config_parameter_column(df, parameter, parameter_function, age_bins)
@@ -176,6 +179,10 @@ def replaceParameters(df, Ki_i, sample_nr, emodl_template, scen_num):
         raise ValueError("Not all placeholders have been replaced in the template emodl file. "
                          f"Remaining placeholders: {remaining_placeholders}")
     fin.close()
+    remaining_placeholders = re.findall(r'@\w+@', data)
+    if remaining_placeholders:
+        raise ValueError("Not all placeholders have been replaced in the template emodl file. "
+                         f"Remaining placeholders: {remaining_placeholders}")
     fin = open(os.path.join(temp_dir, f"simulation_{scen_num}.emodl"), "wt")
     fin.write(data)
     fin.close()

tests/README.md

@@ -0,0 +1,2 @@
+To run unit tests, run the following on the command line:
+```python -m pytest tests```

tests/test_runScenarios.py

@@ -0,0 +1,69 @@
+from datetime import datetime
+
+import pandas as pd
+import pytest
+
+from runScenarios import add_config_parameter_column
+
+
+@pytest.fixture
+def original_df():
+    return pd.DataFrame({'sample_number': [1, 2, 3, 4, 5]})
+
+
+def test_add_config_parameter_column__int(original_df):
+    new_df = add_config_parameter_column(original_df, "new_column", 10)
+    assert new_df.shape == (5, 2)
+    assert "new_column" in new_df.columns
+    assert set(new_df["new_column"]) == set([10])
+
+
+def test_add_config_parameter_column__matrix(original_df):
+    f = {'matrix': [[9, 8], [7, 6]]}
+    new_df = add_config_parameter_column(original_df, "new_column", f)
+    assert new_df.shape == (5, 5)
+    assert all(new_cols in new_df.columns
+               for new_cols in ["new_column1_1", "new_column1_2", "new_column2_1", "new_column2_2"])
+    assert set(new_df["new_column1_1"]) == set([9])
+    assert set(new_df["new_column1_2"]) == set([8])
+    assert set(new_df["new_column2_1"]) == set([7])
+    assert set(new_df["new_column2_2"]) == set([6])
+
+
+def test_add_config_parameter_column__random_uniform(original_df):
+    f = {'np.random': 'uniform', 'function_kwargs': {'low': 5, 'high': 6}}
+    new_df = add_config_parameter_column(original_df, "new_column", f)
+    assert new_df.shape == (5, 2)
+    assert "new_column" in new_df.columns
+    assert all((new_df["new_column"] >= 5) & (new_df["new_column"] <= 6))
+
+
+def test_add_config_parameter_column__datetotimestep():
+    df = pd.DataFrame({'sample_number': [1, 2, 3, 4, 5],
+                       'startdate': [datetime(2020, 2, 20)]*5})
+    f = {'custom_function': 'DateToTimestep',
+         'function_kwargs': {'dates': datetime(2020, 3, 1), 'startdate_col': 'startdate'}}
+    new_df = add_config_parameter_column(df, "new_column", f)
+    assert new_df.shape == (5, 3)
+    assert "new_column" in new_df.columns
+    print(new_df)
+    assert set(new_df["new_column"]) == set([10])
+
+
+def test_add_config_parameter_column__subtract():
+    df = pd.DataFrame({'sample_number': [1, 2, 3, 4, 5],
+                       'col1': [2, 4, 6, 8, 10],
+                       'col2': [1, 3, 5, 7, 9]})
+    f = {'custom_function': 'subtract',
+         'function_kwargs': {'x1': 'col1', 'x2': 'col2'}}
+    new_df = add_config_parameter_column(df, "new_column", f)
+    assert new_df.shape == (5, 4)
+    assert "new_column" in new_df.columns
+    assert set(new_df["new_column"]) == set([1])
+
+
+def test_add_config_parameter_column__error():
+    f = {'weird_function': {}}
+    with pytest.raises(ValueError) as e:
+        add_config_parameter_column(pd.DataFrame, "new_column", f)
+    assert "Unknown type of parameter" in str(e.value)