renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json

018df467 · CI-bot · renku 0.10.4 · 93fd95a3 · 018df467 · 018df467
Commit 018df467 authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/f951cb51a03f4002b77aea8eb00cd432.cwl
+++ b/.renku/workflow/f951cb51a03f4002b77aea8eb00cd432.cwl
+class: Workflow
+cwlVersion: v1.0
+hints: []
+inputs:
+  input_1:
+    default: states-daily.json
+    streamable: false
+    type: string
+  input_2:
+    default: out_folder
+    streamable: false
+    type: string
+  input_3:
+    default: data/covidtracking
+    streamable: false
+    type: string
+  input_4:
+    default:
+      class: File
+      path: ../../notebooks/process/download-covidtracking-data.ipynb
+    streamable: false
+    type: File
+  input_5:
+    default: runs/download-covidtracking-data.runs.ipynb
+    streamable: false
+    type: string
+  input_6:
+    default: states-metadata.json
+    streamable: false
+    type: string
+outputs:
+  output_1:
+    outputSource: step_2/output_1
+    streamable: false
+    type: Directory
+  output_2:
+    outputSource: step_2/output_0
+    streamable: false
+    type: File
+requirements: []
+steps:
+  step_1:
+    in:
+      filename: input_1
+      input_directory: step_2/output_1
+    out:
+    - output_file
+    run:
+      arguments: []
+      baseCommand:
+      - 'true'
+      class: CommandLineTool
+      cwlVersion: v1.0
+      hints: []
+      inputs:
+        filename:
+          default: states-daily.json
+          streamable: false
+          type: string
+        input_directory:
+          streamable: false
+          type: Directory
+      outputs:
+        output_file:
+          outputBinding:
+            glob: $(inputs.filename)
+          streamable: false
+          type: File
+      permanentFailCodes: []
+      requirements:
+      - &id001
+        class: InlineJavascriptRequirement
+      - &id002
+        class: InitialWorkDirRequirement
+        listing: $(inputs.input_directory.listing)
+      successCodes: []
+      temporaryFailCodes: []
+  step_2:
+    in:
+      input_1: input_2
+      input_2: input_3
+      input_3: input_4
+      input_4: input_5
+    out:
+    - output_0
+    - output_1
+    run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl
+  step_3:
+    in:
+      filename: input_6
+      input_directory: step_2/output_1
+    out:
+    - output_file
+    run:
+      arguments: []
+      baseCommand:
+      - 'true'
+      class: CommandLineTool
+      cwlVersion: v1.0
+      hints: []
+      inputs:
+        filename:
+          default: states-metadata.json
+          streamable: false
+          type: string
+        input_directory:
+          streamable: false
+          type: Directory
+      outputs:
+        output_file:
+          outputBinding:
+            glob: $(inputs.filename)
+          streamable: false
+          type: File
+      permanentFailCodes: []
+      requirements:
+      - *id001
+      - *id002
+      successCodes: []
+      temporaryFailCodes: []
--- a/data/covidtracking/states-daily.json
+++ b/data/covidtracking/states-daily.json
--- a/data/covidtracking/states-metadata.json
+++ b/data/covidtracking/states-metadata.json
--- a/runs/download-covidtracking-data.runs.ipynb
+++ b/runs/download-covidtracking-data.runs.ipynb
@@ -5,10 +5,10 @@
   "execution_count": 1,
   "metadata": {
    "papermill": {
-     "duration": 2.730835,
+     "duration": 0.621814,
-     "end_time": "2020-06-13T07:28:02.637600",
+     "end_time": "2020-06-14T07:27:14.855051",
     "exception": false,
-     "start_time": "2020-06-13T07:27:59.906765",
+     "start_time": "2020-06-14T07:27:14.233237",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.023611,
+     "duration": 0.018981,
-     "end_time": "2020-06-13T07:28:02.677718",
+     "end_time": "2020-06-14T07:27:14.888365",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.654107",
+     "start_time": "2020-06-14T07:27:14.869384",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.021893,
+     "duration": 0.021448,
-     "end_time": "2020-06-13T07:28:02.707614",
+     "end_time": "2020-06-14T07:27:14.917138",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.685721",
+     "start_time": "2020-06-14T07:27:14.895690",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/hp7yy9gl/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/cvuzf22v/notebooks/process/download-covidtracking-data.ipynb\"\n",
    "PAPERMILL_OUTPUT_PATH = \"runs/download-covidtracking-data.runs.ipynb\"\n",
    "out_folder = \"data/covidtracking\"\n"
   ]
@@ -68,10 +68,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.007534,
+     "duration": 0.007905,
-     "end_time": "2020-06-13T07:28:02.722850",
+     "end_time": "2020-06-14T07:27:14.934091",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.715316",
+     "start_time": "2020-06-14T07:27:14.926186",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.187654,
+     "duration": 0.451351,
-     "end_time": "2020-06-13T07:28:02.916908",
+     "end_time": "2020-06-14T07:27:15.392017",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.729254",
+     "start_time": "2020-06-14T07:27:14.940666",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.029484,
+     "duration": 0.02628,
-     "end_time": "2020-06-13T07:28:02.961082",
+     "end_time": "2020-06-14T07:27:15.433480",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.931598",
+     "start_time": "2020-06-14T07:27:15.407200",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.085225,
+     "duration": 0.069666,
-     "end_time": "2020-06-13T07:28:03.059893",
+     "end_time": "2020-06-14T07:27:15.514934",
     "exception": false,
-     "start_time": "2020-06-13T07:28:02.974668",
+     "start_time": "2020-06-14T07:27:15.445268",
     "status": "completed"
    },
    "tags": []
@@ -249,10 +249,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.007888,
+     "duration": 0.007442,
-     "end_time": "2020-06-13T07:28:03.082783",
+     "end_time": "2020-06-14T07:27:15.534008",
     "exception": false,
-     "start_time": "2020-06-13T07:28:03.074895",
+     "start_time": "2020-06-14T07:27:15.526566",
     "status": "completed"
    },
    "tags": []
@@ -266,10 +266,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.190354,
+     "duration": 0.354225,
-     "end_time": "2020-06-13T07:28:03.280228",
+     "end_time": "2020-06-14T07:27:15.894431",
     "exception": false,
-     "start_time": "2020-06-13T07:28:03.089874",
+     "start_time": "2020-06-14T07:27:15.540206",
     "status": "completed"
    },
    "tags": []
@@ -286,10 +286,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.019117,
+     "duration": 0.031888,
-     "end_time": "2020-06-13T07:28:03.308684",
+     "end_time": "2020-06-14T07:27:15.940476",
     "exception": false,
-     "start_time": "2020-06-13T07:28:03.289567",
+     "start_time": "2020-06-14T07:27:15.908588",
     "status": "completed"
    },
    "tags": []
@@ -308,10 +308,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.201007,
+     "duration": 0.205405,
-     "end_time": "2020-06-13T07:28:03.516063",
+     "end_time": "2020-06-14T07:27:16.160975",
     "exception": false,
-     "start_time": "2020-06-13T07:28:03.315056",
+     "start_time": "2020-06-14T07:27:15.955570",
     "status": "completed"
    },
    "tags": []
@@ -321,7 +321,7 @@
     "name": "stdout",
     "output_type": "stream",
     "text": [
-      "5561 data points\n"
+      "5617 data points\n"
     ]
    },
    {
@@ -371,21 +371,21 @@
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
-       "      <td>20200612</td>\n",
+       "      <td>20200613</td>\n",
       "      <td>AK</td>\n",
-       "      <td>625.0</td>\n",
+       "      <td>654.0</td>\n",
-       "      <td>70255.0</td>\n",
+       "      <td>71149.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>15.0</td>\n",
+       "      <td>12.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>3.0</td>\n",
+       "      <td>2.0</td>\n",
       "      <td>...</td>\n",
-       "      <td>70880</td>\n",
+       "      <td>71803</td>\n",
-       "      <td>1</td>\n",
+       "      <td>0</td>\n",
       "      <td>0</td>\n",
-       "      <td>5f1a4d5e430995ef9030e190ec73783f54ae0d34</td>\n",
+       "      <td>fd7214547e7a6c8b8153a5a26cb57fa373d2653e</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -395,21 +395,21 @@
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
-       "      <td>20200612</td>\n",
+       "      <td>20200613</td>\n",
       "      <td>AL</td>\n",
-       "      <td>23710.0</td>\n",
+       "      <td>24601.0</td>\n",
-       "      <td>262675.0</td>\n",
+       "      <td>268651.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>622.0</td>\n",
-       "      <td>2202.0</td>\n",
+       "      <td>2241.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>658.0</td>\n",
+       "      <td>672.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
-       "      <td>286385</td>\n",
+       "      <td>293252</td>\n",
-       "      <td>14</td>\n",
+       "      <td>4</td>\n",
-       "      <td>37</td>\n",
+       "      <td>39</td>\n",
-       "      <td>dcbcb0ffeb87d85d6579b48479e7c00de3059a96</td>\n",
+       "      <td>fbbaf3e55677cc7d1eed550a654b675e362268f7</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -424,20 +424,20 @@
      ],
      "text/plain": [
       "       date state  positive  negative  pending  hospitalizedCurrently  \\\n",
-       "0  20200612    AK     625.0   70255.0      NaN                   15.0   \n",
+       "0  20200613    AK     654.0   71149.0      NaN                   12.0   \n",
-       "1  20200612    AL   23710.0  262675.0      NaN                  622.0   \n",
+       "1  20200613    AL   24601.0  268651.0      NaN                  622.0   \n",
       "\n",
       "   hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \\\n",
       "0                     NaN             NaN              NaN   \n",
-       "1                  2202.0             NaN            658.0   \n",
+       "1                  2241.0             NaN            672.0   \n",
       "\n",
       "   onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \\\n",
-       "0                    3.0  ...   70880              1                    0   \n",
+       "0                    2.0  ...   71803              0                    0   \n",
-       "1                    NaN  ...  286385             14                   37   \n",
+       "1                    NaN  ...  293252              4                   39   \n",
       "\n",
       "                                       hash commercialScore  \\\n",
-       "0  5f1a4d5e430995ef9030e190ec73783f54ae0d34               0   \n",
+       "0  fd7214547e7a6c8b8153a5a26cb57fa373d2653e               0   \n",
-       "1  dcbcb0ffeb87d85d6579b48479e7c00de3059a96               0   \n",
+       "1  fbbaf3e55677cc7d1eed550a654b675e362268f7               0   \n",
       "\n",
       "  negativeRegularScore  negativeScore  positiveScore score  grade  \n",
       "0                    0              0              0     0         \n",
@@ -477,18 +477,18 @@
   "version": "3.7.3"
  },
  "papermill": {
-   "duration": 5.591236,
+   "duration": 3.199275,
-   "end_time": "2020-06-13T07:28:03.840975",
+   "end_time": "2020-06-14T07:27:16.489528",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/hp7yy9gl/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/cvuzf22v/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/hp7yy9gl/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/cvuzf22v/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-06-13T07:27:58.249739",
+   "start_time": "2020-06-14T07:27:13.290253",
   "version": "1.1.0"
  }
 },

 %% Cell type:code id: tags:
 ``` python
 import requests
 import os
 import pandas as pd
 ```
 %% Cell type:code id: tags:parameters
 ``` python
 out_folder = "../data/covidtracking/"
 PAPERMILL_OUTPUT_PATH = None
 ```
 %% Cell type:code id: tags:injected-parameters
 ``` python
 # Parameters
-PAPERMILL_INPUT_PATH = "/tmp/hp7yy9gl/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/cvuzf22v/notebooks/process/download-covidtracking-data.ipynb"
 PAPERMILL_OUTPUT_PATH = "runs/download-covidtracking-data.runs.ipynb"
 out_folder = "data/covidtracking"
 ```
 %% Cell type:markdown id: tags:
 # Download state metadata
 Download a dataset of URLs for data for each US state and several territories. See [Google Doc](https://docs.google.com/spreadsheets/d/18oVRrHj3c183mHmq3m89_163yuYltLNlOmPerQ18E8w/htmlview?sle=true).
 %% Cell type:code id: tags:
 ``` python
 url = 'http://covidtracking.com/api/states/info'
 r = requests.get(url, allow_redirects=True)
 states_metadata_json = r.content
 ```
 %% Cell type:code id: tags:
 ``` python
 # save the result
 if PAPERMILL_OUTPUT_PATH:
    out_path = os.path.join(out_folder, 'states-metadata.json')
    with open(out_path, 'wb') as f:
        f.write(states_metadata_json)
 ```
 %% Cell type:code id: tags:
 ``` python
 metadata_df = pd.read_json(states_metadata_json)
 print(len(metadata_df), "states and territories have metadata")
 metadata_df.head(2)
 ```
 %% Output
    56 states and territories have metadata
      state                                              notes  \
    0    AK  Total tests are taken from the annotations on ...
    1    AL  Negatives = (Totals - Positives) \nPositives o...
                                             covid19Site  \
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1  https://alpublichealth.maps.arcgis.com/apps/op...
                                    covid19SiteSecondary covid19SiteTertiary  \
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1            https://dph1.adph.state.al.us/covid-19/
               twitter                                     covid19SiteOld  \
    0     @Alaska_DHSS  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1  @alpublichealth  http://www.alabamapublichealth.gov/infectiousd...
          name  fips pui    pum
    0   Alaska     2      False
    1  Alabama     1      False
 %% Cell type:markdown id: tags:
 # Download daily state data
 %% Cell type:code id: tags:
 ``` python
 url = 'https://covidtracking.com/api/states/daily'
 r = requests.get(url, allow_redirects=True)
 states_daily_json = r.content
 ```
 %% Cell type:code id: tags:
 ``` python
 # save the result
 if PAPERMILL_OUTPUT_PATH:
    out_path = os.path.join(out_folder, 'states-daily.json')
    with open(out_path, 'wb') as f:
        f.write(states_daily_json)
 ```
 %% Cell type:code id: tags:
 ``` python
 data_df = pd.read_json(states_daily_json)
 print(len(data_df), "data points")
 data_df.head(2)
 ```
 %% Output
-    5561 data points
+    5617 data points
           date state  positive  negative  pending  hospitalizedCurrently  \
-    0  20200612    AK     625.0   70255.0      NaN                   15.0
+    0  20200613    AK     654.0   71149.0      NaN                   12.0
-    1  20200612    AL   23710.0  262675.0      NaN                  622.0
+    1  20200613    AL   24601.0  268651.0      NaN                  622.0
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                     NaN             NaN              NaN
-    1                  2202.0             NaN            658.0
+    1                  2241.0             NaN            672.0
       onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \
-    0                    3.0  ...   70880              1                    0
+    0                    2.0  ...   71803              0                    0
-    1                    NaN  ...  286385             14                   37
+    1                    NaN  ...  293252              4                   39
                                           hash commercialScore  \
-    0  5f1a4d5e430995ef9030e190ec73783f54ae0d34               0
+    0  fd7214547e7a6c8b8153a5a26cb57fa373d2653e               0
-    1  dcbcb0ffeb87d85d6579b48479e7c00de3059a96               0
+    1  fbbaf3e55677cc7d1eed550a654b675e362268f7               0
      negativeRegularScore  negativeScore  positiveScore score  grade
    0                    0              0              0     0
    1                    0              0              0     0
    [2 rows x 39 columns]