From def418966f59fb64d92344f523b21fca408e96b1 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Fri, 20 Nov 2020 07:42:22 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../f7f923e1a3704e6285fde7b2c1b1cdc2.cwl | 120 +++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 138 +++++++++--------- 4 files changed, 193 insertions(+), 73 deletions(-) create mode 100644 .renku/workflow/f7f923e1a3704e6285fde7b2c1b1cdc2.cwl diff --git a/.renku/workflow/f7f923e1a3704e6285fde7b2c1b1cdc2.cwl b/.renku/workflow/f7f923e1a3704e6285fde7b2c1b1cdc2.cwl new file mode 100644 index 000000000..fca0574fb --- /dev/null +++ b/.renku/workflow/f7f923e1a3704e6285fde7b2c1b1cdc2.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + 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: states-daily.json + streamable: false + type: string +outputs: + output_2: + outputSource: step_2/output_0 + streamable: false + type: File + output_3: + outputSource: step_2/output_1 + streamable: false + type: Directory +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-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_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-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: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 5a4c39b8d..87b89f934 100644 --- a/data/covidtracking/states-daily.json +++ b/data/covidtracking/states-daily.json @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:8b866cf469f888381452245a3b1c3c68cd034b8634323f09cf555ed8d9326565 -size 19927027 +oid sha256:5554dd423ed693475fcbcfba496440aea84018f5f0bd6e4f14acba79b7373f83 +size 20004386 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index b10317d8c..24826d5a3 100644 --- a/data/covidtracking/states-metadata.json +++ b/data/covidtracking/states-metadata.json @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:33e155b3aa51ca99e45360ddb826f65d77e4068eac535de1e85f1d7e3542de2b -size 127157 +oid sha256:23afc771d7e7037493344a6b194cc893a1ebc2bb47f498a6f0026134963c3342 +size 126837 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index d897315a2..f85c24e9c 100644 --- a/runs/download-covidtracking-data.runs.ipynb +++ b/runs/download-covidtracking-data.runs.ipynb @@ -5,10 +5,10 @@ "execution_count": 1, "metadata": { "papermill": { - "duration": 4.896635, - "end_time": "2020-11-19T07:37:54.448471", + "duration": 6.685221, + "end_time": "2020-11-20T07:42:18.592945", "exception": false, - "start_time": "2020-11-19T07:37:49.551836", + "start_time": "2020-11-20T07:42:11.907724", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.023118, - "end_time": "2020-11-19T07:37:54.486846", + "duration": 0.021916, + "end_time": "2020-11-20T07:42:18.631634", "exception": false, - "start_time": "2020-11-19T07:37:54.463728", + "start_time": "2020-11-20T07:42:18.609718", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.021391, - "end_time": "2020-11-19T07:37:54.520861", + "duration": 0.018925, + "end_time": "2020-11-20T07:42:18.666305", "exception": false, - "start_time": "2020-11-19T07:37:54.499470", + "start_time": "2020-11-20T07:42:18.647380", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/oujakx8z/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/4nizuviw/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": 0.006516, - "end_time": "2020-11-19T07:37:54.536383", + "duration": 0.006326, + "end_time": "2020-11-20T07:42:18.679382", "exception": false, - "start_time": "2020-11-19T07:37:54.529867", + "start_time": "2020-11-20T07:42:18.673056", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.782228, - "end_time": "2020-11-19T07:37:55.324696", + "duration": 0.817947, + "end_time": "2020-11-20T07:42:19.513184", "exception": false, - "start_time": "2020-11-19T07:37:54.542468", + "start_time": "2020-11-20T07:42:18.695237", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.02224, - "end_time": "2020-11-19T07:37:55.358732", + "duration": 0.021783, + "end_time": "2020-11-20T07:42:19.545211", "exception": false, - "start_time": "2020-11-19T07:37:55.336492", + "start_time": "2020-11-20T07:42:19.523428", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.096108, - "end_time": "2020-11-19T07:37:55.464558", + "duration": 0.091027, + "end_time": "2020-11-20T07:42:19.645836", "exception": false, - "start_time": "2020-11-19T07:37:55.368450", + "start_time": "2020-11-20T07:42:19.554809", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007767, - "end_time": "2020-11-19T07:37:55.485516", + "duration": 0.010459, + "end_time": "2020-11-20T07:42:19.668290", "exception": false, - "start_time": "2020-11-19T07:37:55.477749", + "start_time": "2020-11-20T07:42:19.657831", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.603939, - "end_time": "2020-11-19T07:37:56.095902", + "duration": 0.670368, + "end_time": "2020-11-20T07:42:20.345522", "exception": false, - "start_time": "2020-11-19T07:37:55.491963", + "start_time": "2020-11-20T07:42:19.675154", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.06203, - "end_time": "2020-11-19T07:37:56.173711", + "duration": 0.072759, + "end_time": "2020-11-20T07:42:20.429591", "exception": false, - "start_time": "2020-11-19T07:37:56.111681", + "start_time": "2020-11-20T07:42:20.356832", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.571345, - "end_time": "2020-11-19T07:37:56.759551", + "duration": 0.844439, + "end_time": "2020-11-20T07:42:21.285058", "exception": false, - "start_time": "2020-11-19T07:37:56.188206", + "start_time": "2020-11-20T07:42:20.440619", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "14617 data points\n" + "14673 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201118</td>\n", + " <td>20201119</td>\n", " <td>AK</td>\n", - " <td>24419.0</td>\n", + " <td>24909.0</td>\n", " <td>NaN</td>\n", - " <td>861139.0</td>\n", + " <td>873890.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>885558.0</td>\n", - " <td>147.0</td>\n", - " <td>584.0</td>\n", + " <td>898799.0</td>\n", + " <td>139.0</td>\n", + " <td>593.0</td>\n", " <td>...</td>\n", - " <td>885558</td>\n", - " <td>0</td>\n", - " <td>22</td>\n", - " <td>959c531cf9b1dd6d6e9852ef8674d6e5cc0cde51</td>\n", + " <td>898799</td>\n", + " <td>1</td>\n", + " <td>9</td>\n", + " <td>dd60caa3c0b65aa928204156ef405a95116f3f03</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>20201118</td>\n", + " <td>20201119</td>\n", " <td>AL</td>\n", - " <td>223486.0</td>\n", - " <td>35863.0</td>\n", - " <td>1294045.0</td>\n", + " <td>225910.0</td>\n", + " <td>36449.0</td>\n", + " <td>1307256.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1481668.0</td>\n", - " <td>1303.0</td>\n", - " <td>23088.0</td>\n", + " <td>1496717.0</td>\n", + " <td>1315.0</td>\n", + " <td>23295.0</td>\n", " <td>...</td>\n", - " <td>1517531</td>\n", - " <td>46</td>\n", - " <td>372</td>\n", - " <td>670bf40671e729deeede1278bfeeb631084588b3</td>\n", + " <td>1533166</td>\n", + " <td>72</td>\n", + " <td>207</td>\n", + " <td>087ae3543d37e84035c1fde2ac39d1cec933e88d</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 20201118 AK 24419.0 NaN 861139.0 NaN \n", - "1 20201118 AL 223486.0 35863.0 1294045.0 NaN \n", + "0 20201119 AK 24909.0 NaN 873890.0 NaN \n", + "1 20201119 AL 225910.0 36449.0 1307256.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 885558.0 147.0 \n", - "1 totalTestsViral 1481668.0 1303.0 \n", + "0 totalTestsViral 898799.0 139.0 \n", + "1 totalTestsViral 1496717.0 1315.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 584.0 ... 885558 0 22 \n", - "1 23088.0 ... 1517531 46 372 \n", + "0 593.0 ... 898799 1 9 \n", + "1 23295.0 ... 1533166 72 207 \n", "\n", " hash commercialScore \\\n", - "0 959c531cf9b1dd6d6e9852ef8674d6e5cc0cde51 0 \n", - "1 670bf40671e729deeede1278bfeeb631084588b3 0 \n", + "0 dd60caa3c0b65aa928204156ef405a95116f3f03 0 \n", + "1 087ae3543d37e84035c1fde2ac39d1cec933e88d 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": 9.392625, - "end_time": "2020-11-19T07:37:57.084090", + "duration": 12.646574, + "end_time": "2020-11-20T07:42:21.715206", "environment_variables": {}, "exception": null, - "input_path": "/tmp/oujakx8z/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/4nizuviw/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/oujakx8z/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/4nizuviw/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-19T07:37:47.691465", + "start_time": "2020-11-20T07:42:09.068632", "version": "1.1.0" } }, -- GitLab