From 3ca155e7837b5f9d81d8c9a56b7851779f43079e Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Sun, 13 Dec 2020 07:37:48 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../4362ea2d18ab490fa0e5d65099217267.cwl | 120 ++++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 130 +++++++++--------- 4 files changed, 189 insertions(+), 69 deletions(-) create mode 100644 .renku/workflow/4362ea2d18ab490fa0e5d65099217267.cwl diff --git a/.renku/workflow/4362ea2d18ab490fa0e5d65099217267.cwl b/.renku/workflow/4362ea2d18ab490fa0e5d65099217267.cwl new file mode 100644 index 000000000..5cbdadca0 --- /dev/null +++ b/.renku/workflow/4362ea2d18ab490fa0e5d65099217267.cwl @@ -0,0 +1,120 @@ +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_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: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index b155be792..3e31bc293 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:e62d080a170bc871e31539c8be8b3cdd3658cbe5d7fcf8a589eee3aefd0209e4 -size 21724952 +oid sha256:6834cb04ee15ed650960ef5b8d6f14b8fb6b801f46fd0d8b7970b1d163750d72 +size 21802399 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 4a863b13d..2d34465ee 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:f2f507c8365cc03aedb8456a10e41bdea80f74fbd6ed14b16746c6968065691b -size 150268 +oid sha256:7354c67941de362dc05fa040ab2f436300cb1d766c501671e16643a447f1ec46 +size 151830 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index ba4c0680b..7a1003c03 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": 6.889652, - "end_time": "2020-12-12T07:38:05.389643", + "duration": 5.223516, + "end_time": "2020-12-13T07:37:44.521551", "exception": false, - "start_time": "2020-12-12T07:37:58.499991", + "start_time": "2020-12-13T07:37:39.298035", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.015644, - "end_time": "2020-12-12T07:38:05.417406", + "duration": 0.020038, + "end_time": "2020-12-13T07:37:44.555668", "exception": false, - "start_time": "2020-12-12T07:38:05.401762", + "start_time": "2020-12-13T07:37:44.535630", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.010775, - "end_time": "2020-12-12T07:38:05.433035", + "duration": 0.016584, + "end_time": "2020-12-13T07:37:44.579810", "exception": false, - "start_time": "2020-12-12T07:38:05.422260", + "start_time": "2020-12-13T07:37:44.563226", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/r6lieysd/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/_2i_io6h/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": 0.004017, - "end_time": "2020-12-12T07:38:05.441138", + "duration": 0.005697, + "end_time": "2020-12-13T07:37:44.591532", "exception": false, - "start_time": "2020-12-12T07:38:05.437121", + "start_time": "2020-12-13T07:37:44.585835", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.343477, - "end_time": "2020-12-12T07:38:05.788838", + "duration": 1.048541, + "end_time": "2020-12-13T07:37:45.645366", "exception": false, - "start_time": "2020-12-12T07:38:05.445361", + "start_time": "2020-12-13T07:37:44.596825", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.022185, - "end_time": "2020-12-12T07:38:05.822402", + "duration": 0.024147, + "end_time": "2020-12-13T07:37:45.682788", "exception": false, - "start_time": "2020-12-12T07:38:05.800217", + "start_time": "2020-12-13T07:37:45.658641", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.071164, - "end_time": "2020-12-12T07:38:05.902998", + "duration": 0.083844, + "end_time": "2020-12-13T07:37:45.779457", "exception": false, - "start_time": "2020-12-12T07:38:05.831834", + "start_time": "2020-12-13T07:37:45.695613", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.006768, - "end_time": "2020-12-12T07:38:05.921404", + "duration": 0.008011, + "end_time": "2020-12-13T07:37:45.801291", "exception": false, - "start_time": "2020-12-12T07:38:05.914636", + "start_time": "2020-12-13T07:37:45.793280", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.644425, - "end_time": "2020-12-12T07:38:06.571920", + "duration": 0.679409, + "end_time": "2020-12-13T07:37:46.487312", "exception": false, - "start_time": "2020-12-12T07:38:05.927495", + "start_time": "2020-12-13T07:37:45.807903", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.050735, - "end_time": "2020-12-12T07:38:06.639504", + "duration": 0.046951, + "end_time": "2020-12-13T07:37:46.549317", "exception": false, - "start_time": "2020-12-12T07:38:06.588769", + "start_time": "2020-12-13T07:37:46.502366", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.614728, - "end_time": "2020-12-12T07:38:07.266668", + "duration": 0.607161, + "end_time": "2020-12-13T07:37:47.168953", "exception": false, - "start_time": "2020-12-12T07:38:06.651940", + "start_time": "2020-12-13T07:37:46.561792", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "15913 data points\n" + "15969 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201211</td>\n", + " <td>20201212</td>\n", " <td>AK</td>\n", - " <td>38584.0</td>\n", + " <td>39101.0</td>\n", " <td>NaN</td>\n", - " <td>1084961.0</td>\n", + " <td>1094317.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1123545.0</td>\n", + " <td>1133418.0</td>\n", " <td>146.0</td>\n", - " <td>869.0</td>\n", + " <td>883.0</td>\n", " <td>...</td>\n", - " <td>1123545</td>\n", - " <td>3</td>\n", - " <td>9</td>\n", - " <td>47f2cb8f9f01019549b85b0450c8c42c15bb3e77</td>\n", + " <td>1133418</td>\n", + " <td>18</td>\n", + " <td>14</td>\n", + " <td>a2d18ec064bf83b8ac5c9f93a2e0e38827d0865b</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>20201211</td>\n", + " <td>20201212</td>\n", " <td>AL</td>\n", - " <td>288775.0</td>\n", - " <td>50896.0</td>\n", + " <td>292841.0</td>\n", + " <td>52306.0</td>\n", " <td>1458560.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsPeopleViral</td>\n", " <td>1696439.0</td>\n", - " <td>2111.0</td>\n", + " <td>2161.0</td>\n", " <td>28146.0</td>\n", " <td>...</td>\n", - " <td>1747335</td>\n", - " <td>52</td>\n", - " <td>528</td>\n", - " <td>3bf1ccc167c9d139e7506942ea8f53c42bdeeab7</td>\n", + " <td>1751401</td>\n", + " <td>16</td>\n", + " <td>0</td>\n", + " <td>660f90fe723b05c91856029bfd46d43423c107e2</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 20201211 AK 38584.0 NaN 1084961.0 NaN \n", - "1 20201211 AL 288775.0 50896.0 1458560.0 NaN \n", + "0 20201212 AK 39101.0 NaN 1094317.0 NaN \n", + "1 20201212 AL 292841.0 52306.0 1458560.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 1123545.0 146.0 \n", - "1 totalTestsPeopleViral 1696439.0 2111.0 \n", + "0 totalTestsViral 1133418.0 146.0 \n", + "1 totalTestsPeopleViral 1696439.0 2161.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 869.0 ... 1123545 3 9 \n", - "1 28146.0 ... 1747335 52 528 \n", + "0 883.0 ... 1133418 18 14 \n", + "1 28146.0 ... 1751401 16 0 \n", "\n", " hash commercialScore \\\n", - "0 47f2cb8f9f01019549b85b0450c8c42c15bb3e77 0 \n", - "1 3bf1ccc167c9d139e7506942ea8f53c42bdeeab7 0 \n", + "0 a2d18ec064bf83b8ac5c9f93a2e0e38827d0865b 0 \n", + "1 660f90fe723b05c91856029bfd46d43423c107e2 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": 10.7272, - "end_time": "2020-12-12T07:38:07.588984", + "duration": 9.926778, + "end_time": "2020-12-13T07:37:47.490053", "environment_variables": {}, "exception": null, - "input_path": "/tmp/r6lieysd/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/_2i_io6h/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/r6lieysd/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/_2i_io6h/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-12-12T07:37:56.861784", + "start_time": "2020-12-13T07:37:37.563275", "version": "1.1.0" } }, -- GitLab