Refactor JHU smoother to use util:

* remove local smoothing module
* add tests
* update run.py

Author: dshemetov

_delphi_utils_python/delphi_utils/__init__.py

@@ -11,5 +11,6 @@
 from .geomap import GeoMapper
 from .smooth import Smoother
 from .signal import add_prefix, public_signal
+from .smooth import Smoother
 
 __version__ = "0.1.0"

jhu/delphi_jhu/run.py

@@ -6,26 +6,22 @@
 """
 from datetime import datetime
 from itertools import product
-from functools import partial
 
 import numpy as np
+import pandas as pd
 from delphi_utils import (
-    read_params,
-    create_export_csv,
-    S3ArchiveDiffer,
+    read_params, 
+    create_export_csv, 
+    S3ArchiveDiffer, 
+    Smoother, 
+    GeoMapper
 )
 
-from delphi_utils import GeoMapper
 from .geo import geo_map
 from .pull import pull_jhu_data
-from .smooth import (
-    identity,
-    kday_moving_average,
-)
 
 
 # global constants
-seven_day_moving_average = partial(kday_moving_average, k=7)
 METRICS = [
     "deaths",
     "confirmed",
@@ -54,8 +50,8 @@
 #     "cumulative_prop":      ("cumul_prop", False),
 # }
 SMOOTHERS_MAP = {
-    "unsmoothed":           (identity, ''),
-    "seven_day_average":    (seven_day_moving_average, '7dav_'),
+    "unsmoothed":           (Smoother("identity").smooth, ''),
+    "seven_day_average":    (Smoother("moving_average", window_length=7).smooth, '7dav_'),
 }
 GEO_RESOLUTIONS = [
     "county",

jhu/delphi_jhu/smooth.py

@@ -4,7 +4,6 @@
 from os.path import join
 
 import pandas as pd
-from delphi_jhu.run import run_module
 
 
 class TestRun:

jhu/tests/test_run.py

@@ -1,34 +1,28 @@
 import pytest
 
-from os import listdir
 from os.path import join
 
 import numpy as np
 import pandas as pd
-from delphi_jhu.run import run_module
+
 
 class TestSmooth:
     def test_output_files_smoothed(self, run_as_module):
-
-        dates = [str(x) for x in range(20200303, 20200310)]
+        dates = [str(x) for x in range(20200304, 20200311)]
 
         smoothed = pd.read_csv(
-            join("./receiving",
-                f"{dates[-1]}_state_confirmed_7dav_cumulative_num.csv")
+            join("receiving", f"{dates[-1]}_state_confirmed_7dav_cumulative_num.csv")
         )
 
-        # Build a dataframe out of the individual day files
-        raw = pd.concat([
-            pd.read_csv(
-                join("./receiving",
-                    f"{date}_state_confirmed_cumulative_num.csv")
-            ) for date in dates
-        ])
-        # Compute the mean across the time values; order doesn't matter 
-        # this corresponds to the smoothed value on the last day 
-        # 2020-03-10
-        raw = raw.groupby('geo_id')['val'].mean()
-        
-        df = pd.merge(smoothed, raw, on='geo_id', suffixes=('_smoothed', '_raw'))
-        assert np.allclose(df['val_smoothed'].values, df['val_raw'].values)
+        raw = pd.concat(
+            [
+                pd.read_csv(
+                    join("receiving", f"{date}_state_confirmed_cumulative_num.csv")
+                )
+                for date in dates
+            ]
+        )
 
+        raw = raw.groupby("geo_id")["val"].mean()
+        df = pd.merge(smoothed, raw, on="geo_id", suffixes=("_smoothed", "_raw"))
+        assert np.allclose(df["val_smoothed"].values, df["val_raw"].values)