Merge branch 'master' into prescriptor-assignment

Author: ofrancon

Predictor_submission_template.ipynb

@@ -297,7 +297,7 @@
     "    :param end_date: day on which to stop making predictions, as a string, format YYYY-MM-DDD\n",
     "    :param path_to_ips_file: path to a csv file containing the intervention plans between inception date (Jan 1 2020)\n",
     "     and end_date, for the countries and regions for which a prediction is needed\n",
-    "    :param output_file_path: path to file to which to save the the predictions\n",
+    "    :param output_file_path: path to file to save the predictions to\n",
     "    :return: Nothing. Saves the generated predictions to an output_file_path CSV file\n",
     "    with columns \"CountryName,RegionName,Date,PredictedDailyNewCases\"\n",
     "    \"\"\"\n",
@@ -321,7 +321,8 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Display predictions"
+    "## Display predictions\n",
+    "If prediction worked ok, it generated the `output_file` file, that we can read like that:"
    ]
   },
   {
@@ -330,9 +331,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# If prediction worked ok, it generated the following file:\n",
-    "\n",
-    "# That we can read like this:\n",
     "prediction_output_df = pd.read_csv(output_file,\n",
     "                                   parse_dates=['Date'],\n",
     "                                   encoding=\"ISO-8859-1\")\n",

codefresh.yml

@@ -1,6 +1,13 @@
 # Copyright 2020 (c) Cognizant Digital Business, Evolutionary AI. All rights reserved. Issued under the Apache 2.0 License.
 version: '1.0'
 steps:
+  start_clean:
+    title: Clean up CodeFresh volume
+    # Images come from hierarchy at: https://github.com/docker-library/python
+    image: python:3.6-slim
+    commands:
+      - rm -rf '${{CF_REPO_NAME}}'
+
   clone_main_repo:
     type: git-clone
     title: "Clone repo"
@@ -23,3 +30,9 @@ steps:
     working_directory: ${{CF_REPO_NAME}}
     commands:
       - nosetests -v
+
+  end_clean:
+    title: Clean up CodeFresh volume
+    image: python:3.6-slim
+    commands:
+      - rm -rf '${{CF_REPO_NAME}}'

examples/predictors/linear/predict.py

@@ -44,7 +44,7 @@ def predict(start_date: str,
     :param end_date: day on which to stop making predictions, as a string, format YYYY-MM-DDD
     :param path_to_ips_file: path to a csv file containing the intervention plans between inception date (Jan 1 2020)
      and end_date, for the countries and regions for which a prediction is needed
-    :param output_file_path: path to file to which to save the the predictions
+    :param output_file_path: path to file to save the predictions to
     :return: Nothing. Saves the generated predictions to an output_file_path CSV file
     with columns "CountryName,RegionName,Date,PredictedDailyNewCases"
     """

examples/predictors/lstm/predict.py

@@ -30,7 +30,7 @@ def predict(start_date: str,
     :param end_date: day on which to stop making predictions, as a string, format YYYY-MM-DDD
     :param path_to_ips_file: path to a csv file containing the intervention plans between inception date (Jan 1 2020)
      and end_date, for the countries and regions for which a prediction is needed
-    :param output_file_path: path to file to which to save the the predictions
+    :param output_file_path: path to file to save the predictions to
     :return: Nothing. Saves the generated predictions to an output_file_path CSV file
     with columns "CountryName,RegionName,Date,PredictedDailyNewCases"
     """

examples/predictors/lstm/tests/test_xprize_predictor.py

@@ -20,14 +20,12 @@
 END_DATE = "2020-08-04"
 
 
-class TestMultiplicativeEvaluator(unittest.TestCase):
+class TestXPrizePredictor(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
         # Download and cache the raw data
-        latest = True
-        if not os.path.exists(DATA_FILE) or latest:
-            urllib.request.urlretrieve(DATA_URL, DATA_FILE)
+        urllib.request.urlretrieve(DATA_URL, DATA_FILE)
 
     def test_predict(self):
         predictor = XPrizePredictor(PREDICTOR_WEIGHTS, DATA_FILE, CUTOFF_DATE)

examples/predictors/lstm/xprize_predictor.py

@@ -3,6 +3,9 @@
 import os
 import urllib.request
 
+# Suppress noisy Tensorflow debug logging
+os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
+
 # noinspection PyPep8Naming
 import keras.backend as K
 import numpy as np

predict.py

@@ -14,7 +14,7 @@ def predict(start_date: str,
     :param end_date: day on which to stop making predictions, as a string, format YYYY-MM-DDD
     :param path_to_ips_file: path to a csv file containing the intervention plans between inception date (Jan 1 2020)
      and end_date, for the countries and regions for which a prediction is needed
-    :param output_file_path: path to file to which to save the the predictions
+    :param output_file_path: path to file to save the predictions to
     :return: Nothing. Saves the generated predictions to an output_file_path CSV file
     with columns "CountryName,RegionName,Date,PredictedDailyNewCases"
     """

robojudge.ipynb

@@ -163,45 +163,6 @@
     "# actual_npis_df.to_csv(npis_file_name, index=False)"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "hist_end_date_str = \"2020-09-30\"\n",
-    "hist_end_date = pd.to_datetime(hist_end_date_str, format='%Y-%m-%d')"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "historical_npis_df = npis_df[npis_df.Date <= hist_end_date]\n",
-    "historical_npis_df.sample(3)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# hist_file_name = \"validation/data/\" + hist_end_date.strftime('%Y-%m-%d') + \"_historical_ip.csv\"\n",
-    "# # hist_file_name"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# historical_npis_df.to_csv(hist_file_name, index=False)"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},

validation/scenario_generator.py

@@ -84,7 +84,7 @@ def generate_scenario(start_date_str, end_date_str, raw_df, countries=None, scen
         if start_date < INCEPTION_DATE:
             raise ValueError(f"start_date {start_date} must be on or after inception date {INCEPTION_DATE}")
 
-    ips_df = raw_df[ID_COLS + NPI_COLUMNS]
+    ips_df = raw_df[ID_COLS + NPI_COLUMNS].copy()
 
     # Filter on countries
     if countries: