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

fa683785 · CI-bot · renku 0.10.4 · 4be6c119 · fa683785 · fa683785
Commit fa683785 authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/ff585315562b46b585b1de35c7438d42.cwl
+++ b/.renku/workflow/ff585315562b46b585b1de35c7438d42.cwl
+class: Workflow
+cwlVersion: v1.0
+hints: []
+inputs:
+  input_1:
+    default: states-metadata.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-daily.json
+    streamable: false
+    type: string
+outputs:
+  output_0:
+    outputSource: step_2/output_1
+    streamable: false
+    type: Directory
+  output_1:
+    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-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
+        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_1
+    - output_0
+    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-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
+      - *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": 2.95574,
-     "end_time": "2020-06-16T07:27:51.724896",
+     "duration": 1.866113,
+     "end_time": "2020-06-17T07:27:53.277271",
     "exception": false,
-     "start_time": "2020-06-16T07:27:48.769156",
+     "start_time": "2020-06-17T07:27:51.411158",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.025111,
-     "end_time": "2020-06-16T07:27:51.765424",
+     "duration": 0.025595,
+     "end_time": "2020-06-17T07:27:53.317204",
     "exception": false,
-     "start_time": "2020-06-16T07:27:51.740313",
+     "start_time": "2020-06-17T07:27:53.291609",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.021459,
-     "end_time": "2020-06-16T07:27:51.798841",
+     "duration": 0.026336,
+     "end_time": "2020-06-17T07:27:53.357601",
     "exception": false,
-     "start_time": "2020-06-16T07:27:51.777382",
+     "start_time": "2020-06-17T07:27:53.331265",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/65349t7c/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/um7bcmv4/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": 0.005489,
-     "end_time": "2020-06-16T07:27:51.811094",
+     "duration": 0.007976,
+     "end_time": "2020-06-17T07:27:53.379933",
     "exception": false,
-     "start_time": "2020-06-16T07:27:51.805605",
+     "start_time": "2020-06-17T07:27:53.371957",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.178472,
-     "end_time": "2020-06-16T07:27:51.995372",
+     "duration": 0.108755,
+     "end_time": "2020-06-17T07:27:53.495212",
     "exception": false,
-     "start_time": "2020-06-16T07:27:51.816900",
+     "start_time": "2020-06-17T07:27:53.386457",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.024849,
-     "end_time": "2020-06-16T07:27:52.034982",
+     "duration": 0.026513,
+     "end_time": "2020-06-17T07:27:53.537059",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.010133",
+     "start_time": "2020-06-17T07:27:53.510546",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.073662,
-     "end_time": "2020-06-16T07:27:52.119682",
+     "duration": 0.074462,
+     "end_time": "2020-06-17T07:27:53.623699",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.046020",
+     "start_time": "2020-06-17T07:27:53.549237",
     "status": "completed"
    },
    "tags": []
@@ -249,10 +249,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.008628,
-     "end_time": "2020-06-16T07:27:52.142661",
+     "duration": 0.008079,
+     "end_time": "2020-06-17T07:27:53.646468",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.134033",
+     "start_time": "2020-06-17T07:27:53.638389",
     "status": "completed"
    },
    "tags": []
@@ -266,10 +266,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.290266,
-     "end_time": "2020-06-16T07:27:52.439950",
+     "duration": 0.194752,
+     "end_time": "2020-06-17T07:27:53.847777",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.149684",
+     "start_time": "2020-06-17T07:27:53.653025",
     "status": "completed"
    },
    "tags": []
@@ -286,10 +286,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.030335,
-     "end_time": "2020-06-16T07:27:52.488533",
+     "duration": 0.036515,
+     "end_time": "2020-06-17T07:27:53.905748",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.458198",
+     "start_time": "2020-06-17T07:27:53.869233",
     "status": "completed"
    },
    "tags": []
@@ -308,10 +308,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.196926,
-     "end_time": "2020-06-16T07:27:52.697767",
+     "duration": 0.239508,
+     "end_time": "2020-06-17T07:27:54.163296",
     "exception": false,
-     "start_time": "2020-06-16T07:27:52.500841",
+     "start_time": "2020-06-17T07:27:53.923788",
     "status": "completed"
    },
    "tags": []
@@ -321,7 +321,7 @@
     "name": "stdout",
     "output_type": "stream",
     "text": [
-      "5729 data points\n"
+      "5785 data points\n"
     ]
    },
    {
@@ -371,21 +371,21 @@
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
-       "      <td>20200615</td>\n",
+       "      <td>20200616</td>\n",
       "      <td>AK</td>\n",
-       "      <td>664.0</td>\n",
-       "      <td>73773.0</td>\n",
+       "      <td>676.0</td>\n",
+       "      <td>75539.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>21.0</td>\n",
+       "      <td>22.0</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>74437</td>\n",
+       "      <td>76215</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
-       "      <td>6b08035ecccc3d7c158bd1ebff8a325714b92a03</td>\n",
+       "      <td>9ce63702a549318053a366a0ef2d97680c2e6bad</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>20200615</td>\n",
+       "      <td>20200616</td>\n",
       "      <td>AL</td>\n",
-       "      <td>26272.0</td>\n",
-       "      <td>276402.0</td>\n",
+       "      <td>26912.0</td>\n",
+       "      <td>278512.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>546.0</td>\n",
-       "      <td>2259.0</td>\n",
+       "      <td>678.0</td>\n",
+       "      <td>2315.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>676.0</td>\n",
+       "      <td>682.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
-       "      <td>302674</td>\n",
-       "      <td>1</td>\n",
-       "      <td>4</td>\n",
-       "      <td>5a2e19c16964661d50217b2850723b167c2255b8</td>\n",
+       "      <td>305424</td>\n",
+       "      <td>11</td>\n",
+       "      <td>56</td>\n",
+       "      <td>e76cb0a1b1984ba8a95f80bd42da175e42add663</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  20200615    AK     664.0   73773.0      NaN                   21.0   \n",
-       "1  20200615    AL   26272.0  276402.0      NaN                  546.0   \n",
+       "0  20200616    AK     676.0   75539.0      NaN                   22.0   \n",
+       "1  20200616    AL   26912.0  278512.0      NaN                  678.0   \n",
       "\n",
       "   hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \\\n",
       "0                     NaN             NaN              NaN   \n",
-       "1                  2259.0             NaN            676.0   \n",
+       "1                  2315.0             NaN            682.0   \n",
       "\n",
       "   onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \\\n",
-       "0                    3.0  ...   74437              0                    0   \n",
-       "1                    NaN  ...  302674              1                    4   \n",
+       "0                    2.0  ...   76215              0                    0   \n",
+       "1                    NaN  ...  305424             11                   56   \n",
       "\n",
       "                                       hash commercialScore  \\\n",
-       "0  6b08035ecccc3d7c158bd1ebff8a325714b92a03               0   \n",
-       "1  5a2e19c16964661d50217b2850723b167c2255b8               0   \n",
+       "0  9ce63702a549318053a366a0ef2d97680c2e6bad               0   \n",
+       "1  e76cb0a1b1984ba8a95f80bd42da175e42add663               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": 5.92134,
-   "end_time": "2020-06-16T07:27:53.024491",
+   "duration": 5.103315,
+   "end_time": "2020-06-17T07:27:54.492913",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/65349t7c/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/um7bcmv4/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/65349t7c/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/um7bcmv4/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-06-16T07:27:47.103151",
+   "start_time": "2020-06-17T07:27:49.389598",
   "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/65349t7c/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/um7bcmv4/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

-    5729 data points
+    5785 data points

           date state  positive  negative  pending  hospitalizedCurrently  \
-    0  20200615    AK     664.0   73773.0      NaN                   21.0
-    1  20200615    AL   26272.0  276402.0      NaN                  546.0
+    0  20200616    AK     676.0   75539.0      NaN                   22.0
+    1  20200616    AL   26912.0  278512.0      NaN                  678.0
    
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                     NaN             NaN              NaN
-    1                  2259.0             NaN            676.0
+    1                  2315.0             NaN            682.0
    
       onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \
-    0                    3.0  ...   74437              0                    0
-    1                    NaN  ...  302674              1                    4
+    0                    2.0  ...   76215              0                    0
+    1                    NaN  ...  305424             11                   56
    
                                           hash commercialScore  \
-    0  6b08035ecccc3d7c158bd1ebff8a325714b92a03               0
-    1  5a2e19c16964661d50217b2850723b167c2255b8               0
+    0  9ce63702a549318053a366a0ef2d97680c2e6bad               0
+    1  e76cb0a1b1984ba8a95f80bd42da175e42add663               0
    
      negativeRegularScore  negativeScore  positiveScore score  grade
    0                    0              0              0     0
    1                    0              0              0     0
    
    [2 rows x 39 columns]