From 201df7a068a49bd211e04ac6bf4e917efdee7fb2 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Wed, 29 Jul 2020 07:32:53 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../a31dda8820e245c6af116fee39caaeba.cwl | 120 ++++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 128 +++++++++--------- 4 files changed, 188 insertions(+), 68 deletions(-) create mode 100644 .renku/workflow/a31dda8820e245c6af116fee39caaeba.cwl diff --git a/.renku/workflow/a31dda8820e245c6af116fee39caaeba.cwl b/.renku/workflow/a31dda8820e245c6af116fee39caaeba.cwl new file mode 100644 index 000000000..764b082e4 --- /dev/null +++ b/.renku/workflow/a31dda8820e245c6af116fee39caaeba.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: states-daily.json + streamable: false + type: string + input_3: + default: out_folder + streamable: false + type: string + input_4: + default: data/covidtracking + streamable: false + type: string + input_5: + default: + class: File + path: ../../notebooks/process/download-covidtracking-data.ipynb + streamable: false + type: File + input_6: + default: runs/download-covidtracking-data.runs.ipynb + streamable: false + type: string +outputs: + output_2: + outputSource: step_3/output_0 + streamable: false + type: File + output_3: + outputSource: step_3/output_1 + streamable: false + type: Directory +requirements: [] +steps: + step_1: + in: + filename: input_1 + 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 + class: InlineJavascriptRequirement + - &id002 + class: InitialWorkDirRequirement + listing: $(inputs.input_directory.listing) + successCodes: [] + temporaryFailCodes: [] + step_2: + in: + filename: input_2 + input_directory: step_3/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: [] + step_3: + in: + input_1: input_3 + input_2: input_4 + input_3: input_5 + input_4: input_6 + out: + - output_0 + - output_1 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index cdba449b4..1555ddfcf 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 sha256:516201d9016280980a6ee820c53937dbf7240419c40a76cef885c806668d569b -size 7501954 +oid sha256:f7073509fa6668976fd62235a4e4b6bf0da964fb1656fff9738e99f4ea6c69b3 +size 7554680 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 09d0cab52..2160d9b58 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 sha256:44e29bf4c35ce79572343dd09de46c8730020a0703289e8e8996dd5ca125bc7e -size 54647 +oid sha256:c00124de21c434a218eed723fb751d1b2b6c4dd294118e95836ff189ebdbc9cd +size 55384 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 937352e5a..4a91fa1c3 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": 4.721152, - "end_time": "2020-07-28T07:33:19.919400", + "duration": 1.634067, + "end_time": "2020-07-29T07:32:51.603280", "exception": false, - "start_time": "2020-07-28T07:33:15.198248", + "start_time": "2020-07-29T07:32:49.969213", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.022697, - "end_time": "2020-07-28T07:33:19.955433", + "duration": 0.027599, + "end_time": "2020-07-29T07:32:51.645061", "exception": false, - "start_time": "2020-07-28T07:33:19.932736", + "start_time": "2020-07-29T07:32:51.617462", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.018764, - "end_time": "2020-07-28T07:33:19.983298", + "duration": 0.022819, + "end_time": "2020-07-29T07:32:51.676504", "exception": false, - "start_time": "2020-07-28T07:33:19.964534", + "start_time": "2020-07-29T07:32:51.653685", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/zzp5_gpn/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/0mrm7jst/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": 0.008371, - "end_time": "2020-07-28T07:33:19.999719", + "duration": 0.009235, + "end_time": "2020-07-29T07:32:51.700436", "exception": false, - "start_time": "2020-07-28T07:33:19.991348", + "start_time": "2020-07-29T07:32:51.691201", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.106491, - "end_time": "2020-07-28T07:33:20.113724", + "duration": 0.082633, + "end_time": "2020-07-29T07:32:51.789179", "exception": false, - "start_time": "2020-07-28T07:33:20.007233", + "start_time": "2020-07-29T07:32:51.706546", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.024516, - "end_time": "2020-07-28T07:33:20.150633", + "duration": 0.026774, + "end_time": "2020-07-29T07:32:51.838420", "exception": false, - "start_time": "2020-07-28T07:33:20.126117", + "start_time": "2020-07-29T07:32:51.811646", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.074387, - "end_time": "2020-07-28T07:33:20.234935", + "duration": 0.09229, + "end_time": "2020-07-29T07:32:51.941010", "exception": false, - "start_time": "2020-07-28T07:33:20.160548", + "start_time": "2020-07-29T07:32:51.848720", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008074, - "end_time": "2020-07-28T07:33:20.251748", + "duration": 0.007278, + "end_time": "2020-07-29T07:32:51.960099", "exception": false, - "start_time": "2020-07-28T07:33:20.243674", + "start_time": "2020-07-29T07:32:51.952821", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.238079, - "end_time": "2020-07-28T07:33:20.498065", + "duration": 0.313721, + "end_time": "2020-07-29T07:32:52.281007", "exception": false, - "start_time": "2020-07-28T07:33:20.259986", + "start_time": "2020-07-29T07:32:51.967286", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.030914, - "end_time": "2020-07-28T07:33:20.543729", + "duration": 0.035029, + "end_time": "2020-07-29T07:32:52.331598", "exception": false, - "start_time": "2020-07-28T07:33:20.512815", + "start_time": "2020-07-29T07:32:52.296569", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.314232, - "end_time": "2020-07-28T07:33:20.868187", + "duration": 0.368739, + "end_time": "2020-07-29T07:32:52.713521", "exception": false, - "start_time": "2020-07-28T07:33:20.553955", + "start_time": "2020-07-29T07:32:52.344782", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "8081 data points\n" + "8137 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200727</td>\n", + " <td>20200728</td>\n", " <td>AK</td>\n", - " <td>3206.0</td>\n", - " <td>204162.0</td>\n", + " <td>3350.0</td>\n", + " <td>210525.0</td>\n", " <td>NaN</td>\n", - " <td>38.0</td>\n", + " <td>44.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", - " <td>3.0</td>\n", + " <td>4.0</td>\n", " <td>...</td>\n", - " <td>207368</td>\n", + " <td>213875</td>\n", " <td>1</td>\n", " <td>0</td>\n", - " <td>de28ff6d7d2874c791219904c2fa81f83142a78f</td>\n", + " <td>ae5f917caa2078779c9ac0c581af0b3740885be7</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -395,21 +395,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200727</td>\n", + " <td>20200728</td>\n", " <td>AL</td>\n", - " <td>81115.0</td>\n", - " <td>576844.0</td>\n", + " <td>82366.0</td>\n", + " <td>583033.0</td>\n", " <td>NaN</td>\n", - " <td>1599.0</td>\n", - " <td>9694.0</td>\n", + " <td>1598.0</td>\n", + " <td>9893.0</td>\n", " <td>NaN</td>\n", - " <td>1081.0</td>\n", + " <td>1094.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>657959</td>\n", - " <td>18</td>\n", - " <td>537</td>\n", - " <td>906fddfa88864f88bdf0d2bcb19cf267a406f2e6</td>\n", + " <td>665399</td>\n", + " <td>0</td>\n", + " <td>199</td>\n", + " <td>c1a5bd8f18d98fc9530cd541a50772d21cbfb38a</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -424,20 +424,20 @@ ], "text/plain": [ " date state positive negative pending hospitalizedCurrently \\\n", - "0 20200727 AK 3206.0 204162.0 NaN 38.0 \n", - "1 20200727 AL 81115.0 576844.0 NaN 1599.0 \n", + "0 20200728 AK 3350.0 210525.0 NaN 44.0 \n", + "1 20200728 AL 82366.0 583033.0 NaN 1598.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 9694.0 NaN 1081.0 \n", + "1 9893.0 NaN 1094.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 3.0 ... 207368 1 0 \n", - "1 NaN ... 657959 18 537 \n", + "0 4.0 ... 213875 1 0 \n", + "1 NaN ... 665399 0 199 \n", "\n", " hash commercialScore \\\n", - "0 de28ff6d7d2874c791219904c2fa81f83142a78f 0 \n", - "1 906fddfa88864f88bdf0d2bcb19cf267a406f2e6 0 \n", + "0 ae5f917caa2078779c9ac0c581af0b3740885be7 0 \n", + "1 c1a5bd8f18d98fc9530cd541a50772d21cbfb38a 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -477,18 +477,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 7.539678, - "end_time": "2020-07-28T07:33:21.188443", + "duration": 4.508495, + "end_time": "2020-07-29T07:32:53.140366", "environment_variables": {}, "exception": null, - "input_path": "/tmp/zzp5_gpn/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/0mrm7jst/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/zzp5_gpn/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/0mrm7jst/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-07-28T07:33:13.648765", + "start_time": "2020-07-29T07:32:48.631871", "version": "1.1.0" } }, -- GitLab