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

63ad4c94 · CI-bot · renku 0.10.4 · 244d88cf · 63ad4c94 · 63ad4c94
Commit 63ad4c94 authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/0c0a1f3296e7449eabc111d554d13407.cwl
+++ b/.renku/workflow/0c0a1f3296e7449eabc111d554d13407.cwl
+class: Workflow
+cwlVersion: v1.0
+hints: []
+inputs:
+  input_1:
+    default: out_folder
+    streamable: false
+    type: string
+  input_2:
+    default: data/covidtracking
+    streamable: false
+    type: string
+  input_3:
+    default:
+      class: File
+      path: ../../notebooks/process/download-covidtracking-data.ipynb
+    streamable: false
+    type: File
+  input_4:
+    default: runs/download-covidtracking-data.runs.ipynb
+    streamable: false
+    type: string
+  input_5:
+    default: states-daily.json
+    streamable: false
+    type: string
+  input_6:
+    default: states-metadata.json
+    streamable: false
+    type: string
+outputs:
+  output_0:
+    outputSource: step_1/output_1
+    streamable: false
+    type: Directory
+  output_3:
+    outputSource: step_1/output_0
+    streamable: false
+    type: File
+requirements: []
+steps:
+  step_1:
+    in:
+      input_1: input_1
+      input_2: input_2
+      input_3: input_3
+      input_4: input_4
+    out:
+    - output_0
+    - output_1
+    run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl
+  step_2:
+    in:
+      filename: input_5
+      input_directory: step_1/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_3:
+    in:
+      filename: input_6
+      input_directory: step_1/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": 2.264356,
-     "end_time": "2020-06-08T07:27:00.606454",
+     "duration": 3.351525,
+     "end_time": "2020-06-09T07:28:13.679272",
     "exception": false,
-     "start_time": "2020-06-08T07:26:58.342098",
+     "start_time": "2020-06-09T07:28:10.327747",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.018107,
-     "end_time": "2020-06-08T07:27:00.634246",
+     "duration": 0.028314,
+     "end_time": "2020-06-09T07:28:13.724400",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.616139",
+     "start_time": "2020-06-09T07:28:13.696086",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.019065,
-     "end_time": "2020-06-08T07:27:00.660974",
+     "duration": 0.024488,
+     "end_time": "2020-06-09T07:28:13.763732",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.641909",
+     "start_time": "2020-06-09T07:28:13.739244",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/88sodlj1/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/mfmoq2hy/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": 0.008413,
-     "end_time": "2020-06-08T07:27:00.678512",
+     "duration": 0.00777,
+     "end_time": "2020-06-09T07:28:13.786176",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.670099",
+     "start_time": "2020-06-09T07:28:13.778406",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.173251,
-     "end_time": "2020-06-08T07:27:00.859073",
+     "duration": 0.201725,
+     "end_time": "2020-06-09T07:28:13.993473",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.685822",
+     "start_time": "2020-06-09T07:28:13.791748",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.022117,
-     "end_time": "2020-06-08T07:27:00.892647",
+     "duration": 0.027435,
+     "end_time": "2020-06-09T07:28:14.035753",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.870530",
+     "start_time": "2020-06-09T07:28:14.008318",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.068642,
-     "end_time": "2020-06-08T07:27:00.970730",
+     "duration": 0.072843,
+     "end_time": "2020-06-09T07:28:14.120811",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.902088",
+     "start_time": "2020-06-09T07:28:14.047968",
     "status": "completed"
    },
    "tags": []
@@ -249,10 +249,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.008121,
-     "end_time": "2020-06-08T07:27:00.990910",
+     "duration": 0.008497,
+     "end_time": "2020-06-09T07:28:14.143970",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.982789",
+     "start_time": "2020-06-09T07:28:14.135473",
     "status": "completed"
    },
    "tags": []
@@ -266,10 +266,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.640874,
-     "end_time": "2020-06-08T07:27:01.639961",
+     "duration": 0.388675,
+     "end_time": "2020-06-09T07:28:14.539727",
     "exception": false,
-     "start_time": "2020-06-08T07:27:00.999087",
+     "start_time": "2020-06-09T07:28:14.151052",
     "status": "completed"
    },
    "tags": []
@@ -286,10 +286,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.024747,
-     "end_time": "2020-06-08T07:27:01.676632",
+     "duration": 0.032497,
+     "end_time": "2020-06-09T07:28:14.590724",
     "exception": false,
-     "start_time": "2020-06-08T07:27:01.651885",
+     "start_time": "2020-06-09T07:28:14.558227",
     "status": "completed"
    },
    "tags": []
@@ -308,10 +308,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.207692,
-     "end_time": "2020-06-08T07:27:01.895423",
+     "duration": 0.199498,
+     "end_time": "2020-06-09T07:28:14.803254",
     "exception": false,
-     "start_time": "2020-06-08T07:27:01.687731",
+     "start_time": "2020-06-09T07:28:14.603756",
     "status": "completed"
    },
    "tags": []
@@ -321,7 +321,7 @@
     "name": "stdout",
     "output_type": "stream",
     "text": [
-      "5281 data points\n"
+      "5337 data points\n"
     ]
    },
    {
@@ -371,21 +371,21 @@
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
-       "      <td>20200607</td>\n",
+       "      <td>20200608</td>\n",
       "      <td>AK</td>\n",
-       "      <td>544.0</td>\n",
-       "      <td>64360.0</td>\n",
+       "      <td>563.0</td>\n",
+       "      <td>65349.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>1.0</td>\n",
+       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
-       "      <td>64904</td>\n",
+       "      <td>65912</td>\n",
       "      <td>0</td>\n",
-       "      <td>-48</td>\n",
-       "      <td>62adbd451838656b7df7519e830d6439be0b5877</td>\n",
+       "      <td>0</td>\n",
+       "      <td>06594e7f361ff9a5128a34de7a6f674b27c067ed</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>20200607</td>\n",
+       "      <td>20200608</td>\n",
       "      <td>AL</td>\n",
-       "      <td>20500.0</td>\n",
+       "      <td>20925.0</td>\n",
       "      <td>239066.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>2022.0</td>\n",
+       "      <td>2054.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>615.0</td>\n",
+       "      <td>623.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
-       "      <td>259566</td>\n",
-       "      <td>3</td>\n",
-       "      <td>29</td>\n",
-       "      <td>9040674078ce6afca363f8e95943845a032ab5d6</td>\n",
+       "      <td>259991</td>\n",
+       "      <td>26</td>\n",
+       "      <td>32</td>\n",
+       "      <td>b8596160c71e9a7a19933497105b33a6f96ab6f4</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  20200607    AK     544.0   64360.0      NaN                    7.0   \n",
-       "1  20200607    AL   20500.0  239066.0      NaN                    NaN   \n",
+       "0  20200608    AK     563.0   65349.0      NaN                    7.0   \n",
+       "1  20200608    AL   20925.0  239066.0      NaN                    NaN   \n",
       "\n",
       "   hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \\\n",
       "0                     NaN             NaN              NaN   \n",
-       "1                  2022.0             NaN            615.0   \n",
+       "1                  2054.0             NaN            623.0   \n",
       "\n",
       "   onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \\\n",
-       "0                    1.0  ...   64904              0                  -48   \n",
-       "1                    NaN  ...  259566              3                   29   \n",
+       "0                    0.0  ...   65912              0                    0   \n",
+       "1                    NaN  ...  259991             26                   32   \n",
       "\n",
       "                                       hash commercialScore  \\\n",
-       "0  62adbd451838656b7df7519e830d6439be0b5877               0   \n",
-       "1  9040674078ce6afca363f8e95943845a032ab5d6               0   \n",
+       "0  06594e7f361ff9a5128a34de7a6f674b27c067ed               0   \n",
+       "1  b8596160c71e9a7a19933497105b33a6f96ab6f4               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": 6.083115,
-   "end_time": "2020-06-08T07:27:02.218207",
+   "duration": 6.369763,
+   "end_time": "2020-06-09T07:28:15.128265",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/88sodlj1/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/mfmoq2hy/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/88sodlj1/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/mfmoq2hy/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-06-08T07:26:56.135092",
+   "start_time": "2020-06-09T07:28:08.758502",
   "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/88sodlj1/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/mfmoq2hy/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

-    5281 data points
+    5337 data points

           date state  positive  negative  pending  hospitalizedCurrently  \
-    0  20200607    AK     544.0   64360.0      NaN                    7.0
-    1  20200607    AL   20500.0  239066.0      NaN                    NaN
+    0  20200608    AK     563.0   65349.0      NaN                    7.0
+    1  20200608    AL   20925.0  239066.0      NaN                    NaN
    
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                     NaN             NaN              NaN
-    1                  2022.0             NaN            615.0
+    1                  2054.0             NaN            623.0
    
       onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \
-    0                    1.0  ...   64904              0                  -48
-    1                    NaN  ...  259566              3                   29
+    0                    0.0  ...   65912              0                    0
+    1                    NaN  ...  259991             26                   32
    
                                           hash commercialScore  \
-    0  62adbd451838656b7df7519e830d6439be0b5877               0
-    1  9040674078ce6afca363f8e95943845a032ab5d6               0
+    0  06594e7f361ff9a5128a34de7a6f674b27c067ed               0
+    1  b8596160c71e9a7a19933497105b33a6f96ab6f4               0
    
      negativeRegularScore  negativeScore  positiveScore score  grade
    0                    0              0              0     0
    1                    0              0              0     0
    
    [2 rows x 35 columns]