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

4e4b3d94 · CI-bot · renku 0.10.4 · 6ab88db7 · 4e4b3d94 · 4e4b3d94
Commit 4e4b3d94 authored 4 years ago by CI-bot Committed by renku 0.10.4 4 years ago
--- a/.renku/workflow/a2f2b62410c64ffb9f1d30672a23b704.cwl
+++ b/.renku/workflow/a2f2b62410c64ffb9f1d30672a23b704.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_0:
+    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": 2.110873,
-     "end_time": "2020-06-10T07:27:09.995505",
+     "duration": 1.522695,
+     "end_time": "2020-06-10T11:15:28.287816",
     "exception": false,
-     "start_time": "2020-06-10T07:27:07.884632",
+     "start_time": "2020-06-10T11:15:26.765121",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.025521,
-     "end_time": "2020-06-10T07:27:10.033872",
+     "duration": 0.024235,
+     "end_time": "2020-06-10T11:15:28.321893",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.008351",
+     "start_time": "2020-06-10T11:15:28.297658",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.02199,
-     "end_time": "2020-06-10T07:27:10.066732",
+     "duration": 0.024241,
+     "end_time": "2020-06-10T11:15:28.355400",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.044742",
+     "start_time": "2020-06-10T11:15:28.331159",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/06thd704/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/5m4uhyx2/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": 0.007283,
-     "end_time": "2020-06-10T07:27:10.083591",
+     "duration": 0.008908,
+     "end_time": "2020-06-10T11:15:28.373760",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.076308",
+     "start_time": "2020-06-10T11:15:28.364852",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.078331,
-     "end_time": "2020-06-10T07:27:10.169547",
+     "duration": 0.195406,
+     "end_time": "2020-06-10T11:15:28.578569",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.091216",
+     "start_time": "2020-06-10T11:15:28.383163",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.022376,
-     "end_time": "2020-06-10T07:27:10.201874",
+     "duration": 0.02662,
+     "end_time": "2020-06-10T11:15:28.614679",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.179498",
+     "start_time": "2020-06-10T11:15:28.588059",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.069043,
-     "end_time": "2020-06-10T07:27:10.279776",
+     "duration": 0.085122,
+     "end_time": "2020-06-10T11:15:28.709239",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.210733",
+     "start_time": "2020-06-10T11:15:28.624117",
     "status": "completed"
    },
    "tags": []
@@ -249,10 +249,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.008221,
-     "end_time": "2020-06-10T07:27:10.299036",
+     "duration": 0.009986,
+     "end_time": "2020-06-10T11:15:28.730128",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.290815",
+     "start_time": "2020-06-10T11:15:28.720142",
     "status": "completed"
    },
    "tags": []
@@ -266,10 +266,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.168861,
-     "end_time": "2020-06-10T07:27:10.476935",
+     "duration": 0.593309,
+     "end_time": "2020-06-10T11:15:29.333873",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.308074",
+     "start_time": "2020-06-10T11:15:28.740564",
     "status": "completed"
    },
    "tags": []
@@ -286,10 +286,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.02844,
-     "end_time": "2020-06-10T07:27:10.518166",
+     "duration": 0.029815,
+     "end_time": "2020-06-10T11:15:29.382682",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.489726",
+     "start_time": "2020-06-10T11:15:29.352867",
     "status": "completed"
    },
    "tags": []
@@ -308,10 +308,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.217521,
-     "end_time": "2020-06-10T07:27:10.746480",
+     "duration": 0.221971,
+     "end_time": "2020-06-10T11:15:29.616007",
     "exception": false,
-     "start_time": "2020-06-10T07:27:10.528959",
+     "start_time": "2020-06-10T11:15:29.394036",
     "status": "completed"
    },
    "tags": []
@@ -477,18 +477,18 @@
   "version": "3.7.3"
  },
  "papermill": {
-   "duration": 4.583477,
-   "end_time": "2020-06-10T07:27:11.067718",
+   "duration": 4.292822,
+   "end_time": "2020-06-10T11:15:29.939783",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/06thd704/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/5m4uhyx2/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/06thd704/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/5m4uhyx2/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-06-10T07:27:06.484241",
+   "start_time": "2020-06-10T11:15:25.646961",
   "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/06thd704/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/5m4uhyx2/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

    5393 data points

           date state  positive  negative  pending  hospitalizedCurrently  \
    0  20200609    AK     573.0   66317.0      NaN                   12.0
    1  20200609    AL   21422.0  246317.0      NaN                  600.0
    
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                     NaN             NaN              NaN
    1                  2087.0             NaN            631.0
    
       onVentilatorCurrently  ...  posNeg  deathIncrease hospitalizedIncrease  \
    0                    1.0  ...   66890              1                    0
    1                    NaN  ...  267739             11                   33
    
                                           hash commercialScore  \
    0  9eaf1802a047fc56612f95b6e8221d463088d951               0
    1  efc4ed6259393a9d1a456d534f048bed440a3188               0
    
      negativeRegularScore  negativeScore  positiveScore score  grade
    0                    0              0              0     0
    1                    0              0              0     0
    
    [2 rows x 35 columns]