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

ef05452c · CI-bot · renku 0.10.4 · ad7e3fb4 · ef05452c · ef05452c
Commit ef05452c authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/66f5307ec9194324b160ab1fd466e9d6.cwl
+++ b/.renku/workflow/66f5307ec9194324b160ab1fd466e9d6.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_2:
+    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-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": 3.35233,
+     "duration": 4.627897,
-     "end_time": "2020-06-03T15:45:09.318776",
+     "end_time": "2020-06-04T08:28:04.985732",
     "exception": false,
-     "start_time": "2020-06-03T15:45:05.966446",
+     "start_time": "2020-06-04T08:28:00.357835",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.0259,
+     "duration": 0.022291,
-     "end_time": "2020-06-03T15:45:09.361558",
+     "end_time": "2020-06-04T08:28:05.020399",
     "exception": false,
-     "start_time": "2020-06-03T15:45:09.335658",
+     "start_time": "2020-06-04T08:28:04.998108",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.023503,
+     "duration": 0.022545,
-     "end_time": "2020-06-03T15:45:09.397029",
+     "end_time": "2020-06-04T08:28:05.052282",
     "exception": false,
-     "start_time": "2020-06-03T15:45:09.373526",
+     "start_time": "2020-06-04T08:28:05.029737",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/eqrts4dp/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/m_83rm_4/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": 0.006779,
+     "duration": 0.007006,
-     "end_time": "2020-06-03T15:45:09.411335",
+     "end_time": "2020-06-04T08:28:05.068409",
     "exception": false,
-     "start_time": "2020-06-03T15:45:09.404556",
+     "start_time": "2020-06-04T08:28:05.061403",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.597563,
+     "duration": 0.381514,
-     "end_time": "2020-06-03T15:45:10.014708",
+     "end_time": "2020-06-04T08:28:05.457090",
     "exception": false,
-     "start_time": "2020-06-03T15:45:09.417145",
+     "start_time": "2020-06-04T08:28:05.075576",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.03125,
+     "duration": 0.027088,
-     "end_time": "2020-06-03T15:45:10.062585",
+     "end_time": "2020-06-04T08:28:05.496958",
     "exception": false,
-     "start_time": "2020-06-03T15:45:10.031335",
+     "start_time": "2020-06-04T08:28:05.469870",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.091143,
+     "duration": 0.072587,
-     "end_time": "2020-06-03T15:45:10.164517",
+     "end_time": "2020-06-04T08:28:05.580434",
     "exception": false,
-     "start_time": "2020-06-03T15:45:10.073374",
+     "start_time": "2020-06-04T08:28:05.507847",
     "status": "completed"
    },
    "tags": []
@@ -242,10 +242,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.008537,
+     "duration": 0.008248,
-     "end_time": "2020-06-03T15:45:10.188145",
+     "end_time": "2020-06-04T08:28:05.598828",
     "exception": false,
-     "start_time": "2020-06-03T15:45:10.179608",
+     "start_time": "2020-06-04T08:28:05.590580",
     "status": "completed"
    },
    "tags": []
@@ -259,10 +259,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 6.860154,
+     "duration": 0.154111,
-     "end_time": "2020-06-03T15:45:17.054999",
+     "end_time": "2020-06-04T08:28:05.761833",
     "exception": false,
-     "start_time": "2020-06-03T15:45:10.194845",
+     "start_time": "2020-06-04T08:28:05.607722",
     "status": "completed"
    },
    "tags": []
@@ -279,10 +279,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.036298,
+     "duration": 0.029699,
-     "end_time": "2020-06-03T15:45:17.109685",
+     "end_time": "2020-06-04T08:28:05.803127",
     "exception": false,
-     "start_time": "2020-06-03T15:45:17.073387",
+     "start_time": "2020-06-04T08:28:05.773428",
     "status": "completed"
    },
    "tags": []
@@ -301,10 +301,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.224468,
+     "duration": 0.20856,
-     "end_time": "2020-06-03T15:45:17.346605",
+     "end_time": "2020-06-04T08:28:06.022260",
     "exception": false,
-     "start_time": "2020-06-03T15:45:17.122137",
+     "start_time": "2020-06-04T08:28:05.813700",
     "status": "completed"
    },
    "tags": []
@@ -314,7 +314,7 @@
     "name": "stdout",
     "output_type": "stream",
     "text": [
-      "5001 data points\n"
+      "5057 data points\n"
     ]
    },
    {
@@ -364,21 +364,21 @@
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
-       "      <td>20200602</td>\n",
+       "      <td>20200603</td>\n",
       "      <td>AK</td>\n",
-       "      <td>487.0</td>\n",
+       "      <td>505.0</td>\n",
-       "      <td>55716.0</td>\n",
+       "      <td>57677.0</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>10.0</td>\n",
+       "      <td>11.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
-       "      <td>56203</td>\n",
+       "      <td>58182</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
-       "      <td>8bfc576f3ce0c5438919ab2a672255d0630714ec</td>\n",
+       "      <td>02548c93014e510ca02e0fc92122c299e4aa8e78</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -388,21 +388,21 @@
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
-       "      <td>20200602</td>\n",
+       "      <td>20200603</td>\n",
       "      <td>AL</td>\n",
-       "      <td>18642.0</td>\n",
+       "      <td>18851.0</td>\n",
-       "      <td>205160.0</td>\n",
+       "      <td>212743.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
-       "      <td>1879.0</td>\n",
+       "      <td>1900.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>591.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
-       "      <td>223802</td>\n",
+       "      <td>231594</td>\n",
-       "      <td>5</td>\n",
+       "      <td>2</td>\n",
-       "      <td>23</td>\n",
+       "      <td>21</td>\n",
-       "      <td>e925d317dd6bba774728abaf5db247f3c5983394</td>\n",
+       "      <td>fe5df10e9c8639e15d2ce02d16d678e16f6f6cc7</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
@@ -417,20 +417,20 @@
      ],
      "text/plain": [
       "       date state  positive  negative  pending  hospitalizedCurrently  \\\n",
-       "0  20200602    AK     487.0   55716.0      NaN                   10.0   \n",
+       "0  20200603    AK     505.0   57677.0      NaN                   11.0   \n",
-       "1  20200602    AL   18642.0  205160.0      NaN                    NaN   \n",
+       "1  20200603    AL   18851.0  212743.0      NaN                    NaN   \n",
       "\n",
       "   hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \\\n",
       "0                     NaN             NaN              NaN   \n",
-       "1                  1879.0             NaN            591.0   \n",
+       "1                  1900.0             NaN            591.0   \n",
       "\n",
       "   onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \\\n",
-       "0                    0.0  ...   56203              0                    0   \n",
+       "0                    0.0  ...   58182              0                    0   \n",
-       "1                    NaN  ...  223802              5                   23   \n",
+       "1                    NaN  ...  231594              2                   21   \n",
       "\n",
       "                                       hash commercialScore  \\\n",
-       "0  8bfc576f3ce0c5438919ab2a672255d0630714ec               0   \n",
+       "0  02548c93014e510ca02e0fc92122c299e4aa8e78               0   \n",
-       "1  e925d317dd6bba774728abaf5db247f3c5983394               0   \n",
+       "1  fe5df10e9c8639e15d2ce02d16d678e16f6f6cc7               0   \n",
       "\n",
       "  negativeRegularScore  negativeScore  positiveScore score  grade  \n",
       "0                    0              0              0     0         \n",
@@ -470,18 +470,18 @@
   "version": "3.7.3"
  },
  "papermill": {
-   "duration": 13.234479,
+   "duration": 7.919109,
-   "end_time": "2020-06-03T15:45:17.775676",
+   "end_time": "2020-06-04T08:28:06.447225",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/eqrts4dp/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/m_83rm_4/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/eqrts4dp/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/m_83rm_4/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-06-03T15:45:04.541197",
+   "start_time": "2020-06-04T08:27:58.528116",
   "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/eqrts4dp/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/m_83rm_4/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          twitter  \
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...     @Alaska_DHSS
    1            https://dph1.adph.state.al.us/covid-19/  @alpublichealth
                                          covid19SiteOld     name  fips pui    pum
    0  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...   Alaska     2      False
    1  http://www.alabamapublichealth.gov/infectiousd...  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
-    5001 data points
+    5057 data points
           date state  positive  negative  pending  hospitalizedCurrently  \
-    0  20200602    AK     487.0   55716.0      NaN                   10.0
+    0  20200603    AK     505.0   57677.0      NaN                   11.0
-    1  20200602    AL   18642.0  205160.0      NaN                    NaN
+    1  20200603    AL   18851.0  212743.0      NaN                    NaN
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                     NaN             NaN              NaN
-    1                  1879.0             NaN            591.0
+    1                  1900.0             NaN            591.0
       onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \
-    0                    0.0  ...   56203              0                    0
+    0                    0.0  ...   58182              0                    0
-    1                    NaN  ...  223802              5                   23
+    1                    NaN  ...  231594              2                   21
                                           hash commercialScore  \
-    0  8bfc576f3ce0c5438919ab2a672255d0630714ec               0
+    0  02548c93014e510ca02e0fc92122c299e4aa8e78               0
-    1  e925d317dd6bba774728abaf5db247f3c5983394               0
+    1  fe5df10e9c8639e15d2ce02d16d678e16f6f6cc7               0
      negativeRegularScore  negativeScore  positiveScore score  grade
    0                    0              0              0     0
    1                    0              0              0     0
    [2 rows x 35 columns]