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

53845455 · CI-bot · renku 0.10.4 · 6a0dd6ef · 53845455 · 53845455
Commit 53845455 authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/01af5e5b09a643b296080a4480511bcf.cwl
+++ b/.renku/workflow/01af5e5b09a643b296080a4480511bcf.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-metadata.json
+    streamable: false
+    type: string
+  input_6:
+    default: states-daily.json
+    streamable: false
+    type: string
+outputs:
+  output_0:
+    outputSource: step_1/output_0
+    streamable: false
+    type: File
+  output_3:
+    outputSource: step_1/output_1
+    streamable: false
+    type: Directory
+requirements: []
+steps:
+  step_1:
+    in:
+      input_1: input_1
+      input_2: input_2
+      input_3: input_3
+      input_4: input_4
+    out:
+    - output_1
+    - output_0
+    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-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_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-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": 5.531971,
-     "end_time": "2020-11-16T07:36:41.614756",
+     "duration": 5.032173,
+     "end_time": "2020-11-17T07:42:42.936102",
     "exception": false,
-     "start_time": "2020-11-16T07:36:36.082785",
+     "start_time": "2020-11-17T07:42:37.903929",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.018945,
-     "end_time": "2020-11-16T07:36:41.648956",
+     "duration": 0.029006,
+     "end_time": "2020-11-17T07:42:42.981669",
     "exception": false,
-     "start_time": "2020-11-16T07:36:41.630011",
+     "start_time": "2020-11-17T07:42:42.952663",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.011935,
-     "end_time": "2020-11-16T07:36:41.668692",
+     "duration": 0.026,
+     "end_time": "2020-11-17T07:42:43.021237",
     "exception": false,
-     "start_time": "2020-11-16T07:36:41.656757",
+     "start_time": "2020-11-17T07:42:42.995237",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/8a0rjkgq/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/nu8t6zmk/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": 0.0049,
-     "end_time": "2020-11-16T07:36:41.677826",
+     "duration": 0.007143,
+     "end_time": "2020-11-17T07:42:43.039079",
     "exception": false,
-     "start_time": "2020-11-16T07:36:41.672926",
+     "start_time": "2020-11-17T07:42:43.031936",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.390166,
-     "end_time": "2020-11-16T07:36:42.072559",
+     "duration": 0.399521,
+     "end_time": "2020-11-17T07:42:43.444673",
     "exception": false,
-     "start_time": "2020-11-16T07:36:41.682393",
+     "start_time": "2020-11-17T07:42:43.045152",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.022204,
-     "end_time": "2020-11-16T07:36:42.107795",
+     "duration": 0.028097,
+     "end_time": "2020-11-17T07:42:43.488259",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.085591",
+     "start_time": "2020-11-17T07:42:43.460162",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.064797,
-     "end_time": "2020-11-16T07:36:42.179830",
+     "duration": 0.10648,
+     "end_time": "2020-11-17T07:42:43.604236",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.115033",
+     "start_time": "2020-11-17T07:42:43.497756",
     "status": "completed"
    },
    "tags": []
@@ -284,10 +284,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.00734,
-     "end_time": "2020-11-16T07:36:42.199365",
+     "duration": 0.007903,
+     "end_time": "2020-11-17T07:42:43.625221",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.192025",
+     "start_time": "2020-11-17T07:42:43.617318",
     "status": "completed"
    },
    "tags": []
@@ -301,10 +301,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.441262,
-     "end_time": "2020-11-16T07:36:42.646845",
+     "duration": 0.649363,
+     "end_time": "2020-11-17T07:42:44.280699",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.205583",
+     "start_time": "2020-11-17T07:42:43.631336",
     "status": "completed"
    },
    "tags": []
@@ -321,10 +321,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.167171,
-     "end_time": "2020-11-16T07:36:42.829617",
+     "duration": 0.055423,
+     "end_time": "2020-11-17T07:42:44.355421",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.662446",
+     "start_time": "2020-11-17T07:42:44.299998",
     "status": "completed"
    },
    "tags": []
@@ -343,10 +343,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.568847,
-     "end_time": "2020-11-16T07:36:43.413261",
+     "duration": 0.811442,
+     "end_time": "2020-11-17T07:42:45.181932",
     "exception": false,
-     "start_time": "2020-11-16T07:36:42.844414",
+     "start_time": "2020-11-17T07:42:44.370490",
     "status": "completed"
    },
    "tags": []
@@ -356,7 +356,7 @@
     "name": "stdout",
     "output_type": "stream",
     "text": [
-      "14447 data points\n"
+      "14503 data points\n"
     ]
    },
    {
@@ -406,21 +406,21 @@
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
-       "      <td>20201115</td>\n",
+       "      <td>20201116</td>\n",
       "      <td>AK</td>\n",
-       "      <td>23814.0</td>\n",
+       "      <td>23240.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>843568.0</td>\n",
+       "      <td>849107.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>totalTestsViral</td>\n",
-       "      <td>867382.0</td>\n",
-       "      <td>141.0</td>\n",
-       "      <td>NaN</td>\n",
+       "      <td>872347.0</td>\n",
+       "      <td>143.0</td>\n",
+       "      <td>559.0</td>\n",
       "      <td>...</td>\n",
-       "      <td>867382</td>\n",
-       "      <td>0</td>\n",
+       "      <td>872347</td>\n",
       "      <td>0</td>\n",
-       "      <td>e932ee5bb80cf68522ff67db8d19bd8e6e3f3c1c</td>\n",
+       "      <td>9</td>\n",
+       "      <td>f9555880eaacb39588e3dbdc1822f7fca35ebf82</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -430,21 +430,21 @@
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
-       "      <td>20201115</td>\n",
+       "      <td>20201116</td>\n",
       "      <td>AL</td>\n",
-       "      <td>217822.0</td>\n",
-       "      <td>34331.0</td>\n",
-       "      <td>1275915.0</td>\n",
+       "      <td>219232.0</td>\n",
+       "      <td>34587.0</td>\n",
+       "      <td>1281958.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>totalTestsViral</td>\n",
-       "      <td>1459406.0</td>\n",
-       "      <td>1195.0</td>\n",
-       "      <td>22275.0</td>\n",
+       "      <td>1466603.0</td>\n",
+       "      <td>1262.0</td>\n",
+       "      <td>22716.0</td>\n",
       "      <td>...</td>\n",
-       "      <td>1493737</td>\n",
-       "      <td>2</td>\n",
-       "      <td>0</td>\n",
-       "      <td>bd0aca6268151786d8349818e098840bf0ce6a42</td>\n",
+       "      <td>1501190</td>\n",
+       "      <td>1</td>\n",
+       "      <td>441</td>\n",
+       "      <td>6a3bed8bae24c6fd056a322cdf8bdb75b8e91348</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -459,20 +459,20 @@
      ],
      "text/plain": [
       "       date state  positive  probableCases   negative  pending  \\\n",
-       "0  20201115    AK   23814.0            NaN   843568.0      NaN   \n",
-       "1  20201115    AL  217822.0        34331.0  1275915.0      NaN   \n",
+       "0  20201116    AK   23240.0            NaN   849107.0      NaN   \n",
+       "1  20201116    AL  219232.0        34587.0  1281958.0      NaN   \n",
       "\n",
       "  totalTestResultsSource  totalTestResults  hospitalizedCurrently  \\\n",
-       "0        totalTestsViral          867382.0                  141.0   \n",
-       "1        totalTestsViral         1459406.0                 1195.0   \n",
+       "0        totalTestsViral          872347.0                  143.0   \n",
+       "1        totalTestsViral         1466603.0                 1262.0   \n",
       "\n",
       "   hospitalizedCumulative  ...   posNeg  deathIncrease  hospitalizedIncrease  \\\n",
-       "0                     NaN  ...   867382              0                     0   \n",
-       "1                 22275.0  ...  1493737              2                     0   \n",
+       "0                   559.0  ...   872347              0                     9   \n",
+       "1                 22716.0  ...  1501190              1                   441   \n",
       "\n",
       "                                       hash  commercialScore  \\\n",
-       "0  e932ee5bb80cf68522ff67db8d19bd8e6e3f3c1c                0   \n",
-       "1  bd0aca6268151786d8349818e098840bf0ce6a42                0   \n",
+       "0  f9555880eaacb39588e3dbdc1822f7fca35ebf82                0   \n",
+       "1  6a3bed8bae24c6fd056a322cdf8bdb75b8e91348                0   \n",
       "\n",
       "  negativeRegularScore negativeScore positiveScore score  grade  \n",
       "0                    0             0             0     0         \n",
@@ -512,18 +512,18 @@
   "version": "3.7.3"
  },
  "papermill": {
-   "duration": 9.282661,
-   "end_time": "2020-11-16T07:36:43.734701",
+   "duration": 9.728724,
+   "end_time": "2020-11-17T07:42:45.610567",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/8a0rjkgq/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/nu8t6zmk/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/8a0rjkgq/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/nu8t6zmk/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-11-16T07:36:34.452040",
+   "start_time": "2020-11-17T07:42:35.881843",
   "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/8a0rjkgq/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/nu8t6zmk/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  Alaska combines PCR and antigen tests in the t...
    1    AL  Alabama combines PCR and antigen tests in the ...
    
                                             covid19Site  \
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1  https://alpublichealth.maps.arcgis.com/apps/op...
    
                                    covid19SiteSecondary  \
    0  https://experience.arcgis.com/experience/ed1c8...
    1  https://alpublichealth.maps.arcgis.com/apps/op...
    
                                     covid19SiteTertiary  \
    0  https://alaska-dhss.maps.arcgis.com/apps/opsda...
    1  https://services7.arcgis.com/4RQmZZ0yaZkGR1zy/...
    
                                   covid19SiteQuaternary  \
    0  https://services1.arcgis.com/WzFsmainVTuD5KML/...
    1                                               None
    
                                      covid19SiteQuinary          twitter  \
    0  https://services1.arcgis.com/WzFsmainVTuD5KML/...     @Alaska_DHSS
    1                                               None  @alpublichealth
    
                                          covid19SiteOld  \
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1  http://www.alabamapublichealth.gov/infectiousd...
    
      covidTrackingProjectPreferredTotalTestUnits  \
    0                                   Specimens
    1                               Unclear units
    
      covidTrackingProjectPreferredTotalTestField totalTestResultsField       pui  \
    0                             totalTestsViral     Total Tests (PCR)  All data
    1                             totalTestsViral     Total Tests (PCR)   No data
    
         pum     name  fips
    0  False   Alaska     2
    1  False  Alabama     1

 %% 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

-    14447 data points
+    14503 data points

           date state  positive  probableCases   negative  pending  \
-    0  20201115    AK   23814.0            NaN   843568.0      NaN
-    1  20201115    AL  217822.0        34331.0  1275915.0      NaN
+    0  20201116    AK   23240.0            NaN   849107.0      NaN
+    1  20201116    AL  219232.0        34587.0  1281958.0      NaN
    
      totalTestResultsSource  totalTestResults  hospitalizedCurrently  \
-    0        totalTestsViral          867382.0                  141.0
-    1        totalTestsViral         1459406.0                 1195.0
+    0        totalTestsViral          872347.0                  143.0
+    1        totalTestsViral         1466603.0                 1262.0
    
       hospitalizedCumulative  ...   posNeg  deathIncrease  hospitalizedIncrease  \
-    0                     NaN  ...   867382              0                     0
-    1                 22275.0  ...  1493737              2                     0
+    0                   559.0  ...   872347              0                     9
+    1                 22716.0  ...  1501190              1                   441
    
                                           hash  commercialScore  \
-    0  e932ee5bb80cf68522ff67db8d19bd8e6e3f3c1c                0
-    1  bd0aca6268151786d8349818e098840bf0ce6a42                0
+    0  f9555880eaacb39588e3dbdc1822f7fca35ebf82                0
+    1  6a3bed8bae24c6fd056a322cdf8bdb75b8e91348                0
    
      negativeRegularScore negativeScore positiveScore score  grade
    0                    0             0             0     0
    1                    0             0             0     0
    
    [2 rows x 55 columns]