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

b7bbbf0f · CR (covid cron) · renku 0.10.2 · f7617381 · b7bbbf0f · b7bbbf0f
Commit b7bbbf0f authored 5 years ago by CR (covid cron) Committed by renku 0.10.2 5 years ago
--- a/.renku/workflow/9c0f49623aa94b0d85c3de74aae5f32a.cwl
+++ b/.renku/workflow/9c0f49623aa94b0d85c3de74aae5f32a.cwl
+class: Workflow
+cwlVersion: v1.0
+hints: []
+inputs:
+  input_1:
+    default: states-daily.json
+    streamable: false
+    type: string
+  input_10:
+    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: out_folder
+    streamable: false
+    type: string
+  input_7:
+    default: data/covidtracking
+    streamable: false
+    type: string
+  input_8:
+    default:
+      class: File
+      path: ../../notebooks/process/download-covidtracking-data.ipynb
+    streamable: false
+    type: File
+  input_9:
+    default: runs/download-covidtracking-data.runs.ipynb
+    streamable: false
+    type: string
+outputs:
+  output_1:
+    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_1
+    - output_0
+    run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl
+  step_3:
+    in:
+      input_1: input_6
+      input_2: input_7
+      input_3: input_8
+      input_4: input_9
+    out:
+    - output_1
+    - output_0
+    run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl
+  step_4:
+    in:
+      filename: input_10
+      input_directory: step_3/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": 0.483354,
-     "end_time": "2020-04-04T07:25:10.355198",
+     "duration": 4.118811,
+     "end_time": "2020-04-04T21:04:50.001206",
     "exception": false,
-     "start_time": "2020-04-04T07:25:09.871844",
+     "start_time": "2020-04-04T21:04:45.882395",
     "status": "completed"
    },
    "tags": []
@@ -25,10 +25,10 @@
   "execution_count": 2,
   "metadata": {
    "papermill": {
-     "duration": 0.015684,
-     "end_time": "2020-04-04T07:25:10.380598",
+     "duration": 0.017727,
+     "end_time": "2020-04-04T21:04:50.032092",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.364914",
+     "start_time": "2020-04-04T21:04:50.014365",
     "status": "completed"
    },
    "tags": [
@@ -46,10 +46,10 @@
   "execution_count": 3,
   "metadata": {
    "papermill": {
-     "duration": 0.015223,
-     "end_time": "2020-04-04T07:25:10.402953",
+     "duration": 0.021211,
+     "end_time": "2020-04-04T21:04:50.062863",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.387730",
+     "start_time": "2020-04-04T21:04:50.041652",
     "status": "completed"
    },
    "tags": [
@@ -59,7 +59,7 @@
   "outputs": [],
   "source": [
    "# Parameters\n",
-    "PAPERMILL_INPUT_PATH = \"/tmp/lou3ghqx/notebooks/process/download-covidtracking-data.ipynb\"\n",
+    "PAPERMILL_INPUT_PATH = \"/tmp/cqky7dsw/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": 0.006883,
-     "end_time": "2020-04-04T07:25:10.417247",
+     "duration": 0.008752,
+     "end_time": "2020-04-04T21:04:50.085061",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.410364",
+     "start_time": "2020-04-04T21:04:50.076309",
     "status": "completed"
    },
    "tags": []
@@ -87,10 +87,10 @@
   "execution_count": 4,
   "metadata": {
    "papermill": {
-     "duration": 0.237124,
-     "end_time": "2020-04-04T07:25:10.660579",
+     "duration": 1.082816,
+     "end_time": "2020-04-04T21:04:51.175359",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.423455",
+     "start_time": "2020-04-04T21:04:50.092543",
     "status": "completed"
    },
    "tags": []
@@ -107,10 +107,10 @@
   "execution_count": 5,
   "metadata": {
    "papermill": {
-     "duration": 0.022094,
-     "end_time": "2020-04-04T07:25:10.694146",
+     "duration": 0.021503,
+     "end_time": "2020-04-04T21:04:51.208251",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.672052",
+     "start_time": "2020-04-04T21:04:51.186748",
     "status": "completed"
    },
    "tags": []
@@ -129,10 +129,10 @@
   "execution_count": 6,
   "metadata": {
    "papermill": {
-     "duration": 0.065719,
-     "end_time": "2020-04-04T07:25:10.768853",
+     "duration": 0.069639,
+     "end_time": "2020-04-04T21:04:51.286125",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.703134",
+     "start_time": "2020-04-04T21:04:51.216486",
     "status": "completed"
    },
    "tags": []
@@ -246,10 +246,10 @@
   "cell_type": "markdown",
   "metadata": {
    "papermill": {
-     "duration": 0.007831,
-     "end_time": "2020-04-04T07:25:10.786260",
+     "duration": 0.008089,
+     "end_time": "2020-04-04T21:04:51.303421",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.778429",
+     "start_time": "2020-04-04T21:04:51.295332",
     "status": "completed"
    },
    "tags": []
@@ -263,10 +263,10 @@
   "execution_count": 7,
   "metadata": {
    "papermill": {
-     "duration": 0.215185,
-     "end_time": "2020-04-04T07:25:11.009344",
+     "duration": 0.235648,
+     "end_time": "2020-04-04T21:04:51.547279",
     "exception": false,
-     "start_time": "2020-04-04T07:25:10.794159",
+     "start_time": "2020-04-04T21:04:51.311631",
     "status": "completed"
    },
    "tags": []
@@ -283,10 +283,10 @@
   "execution_count": 8,
   "metadata": {
    "papermill": {
-     "duration": 0.025212,
-     "end_time": "2020-04-04T07:25:11.045222",
+     "duration": 0.023705,
+     "end_time": "2020-04-04T21:04:51.585035",
     "exception": false,
-     "start_time": "2020-04-04T07:25:11.020010",
+     "start_time": "2020-04-04T21:04:51.561330",
     "status": "completed"
    },
    "tags": []
@@ -305,10 +305,10 @@
   "execution_count": 9,
   "metadata": {
    "papermill": {
-     "duration": 0.093239,
-     "end_time": "2020-04-04T07:25:11.151043",
+     "duration": 0.093556,
+     "end_time": "2020-04-04T21:04:51.689668",
     "exception": false,
-     "start_time": "2020-04-04T07:25:11.057804",
+     "start_time": "2020-04-04T21:04:51.596112",
     "status": "completed"
    },
    "tags": []
@@ -474,18 +474,18 @@
   "version": "3.7.3"
  },
  "papermill": {
-   "duration": 2.432226,
-   "end_time": "2020-04-04T07:25:11.472698",
+   "duration": 7.617778,
+   "end_time": "2020-04-04T21:04:52.010906",
   "environment_variables": {},
   "exception": null,
-   "input_path": "/tmp/lou3ghqx/notebooks/process/download-covidtracking-data.ipynb",
+   "input_path": "/tmp/cqky7dsw/notebooks/process/download-covidtracking-data.ipynb",
   "output_path": "runs/download-covidtracking-data.runs.ipynb",
   "parameters": {
-    "PAPERMILL_INPUT_PATH": "/tmp/lou3ghqx/notebooks/process/download-covidtracking-data.ipynb",
+    "PAPERMILL_INPUT_PATH": "/tmp/cqky7dsw/notebooks/process/download-covidtracking-data.ipynb",
    "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
    "out_folder": "data/covidtracking"
   },
-   "start_time": "2020-04-04T07:25:09.040472",
+   "start_time": "2020-04-04T21:04:44.393128",
   "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/lou3ghqx/notebooks/process/download-covidtracking-data.ipynb"
+PAPERMILL_INPUT_PATH = "/tmp/cqky7dsw/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                                     covid19SiteOld  \
    0    AK  http://dhss.alaska.gov/dph/Epi/id/Pages/COVID-...
    1    AL  http://www.alabamapublichealth.gov/infectiousd...
    
                                             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                                               None  @alpublichealth
    
            pui    pum                                              notes  fips  \
    0  All data  False  Total tests are taken from the annotations on ...     2
    1   No data  False  Negatives = (Totals - Positives) \nPositives o...     1
    
          name
    0   Alaska
    1  Alabama

 %% 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

    1597 data points

           date state  positive  negative  pending  hospitalizedCurrently  \
    0  20200403    AK     157.0    5859.0      NaN                    NaN
    1  20200403    AL    1432.0    8187.0      NaN                    NaN
    
       hospitalizedCumulative  inIcuCurrently  inIcuCumulative  \
    0                    15.0             NaN              NaN
    1                     NaN             NaN              NaN
    
       onVentilatorCurrently  ...  hospitalized  total totalTestResults posNeg  \
    0                    NaN  ...          15.0   6016             6016   6016
    1                    NaN  ...           NaN   9619             9619   9619
    
       fips  deathIncrease  hospitalizedIncrease  negativeIncrease  \
    0     2            0.0                   6.0             980.0
    1     1            3.0                   0.0             684.0
    
       positiveIncrease  totalTestResultsIncrease
    0              14.0                     994.0
    1             199.0                     883.0
    
    [2 rows x 25 columns]