From 925756b69c27cff7df779633d5b0598a086784e3 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Wed, 16 Sep 2020 07:35:51 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../ed22b9067a714a60aa63af2ba94b8a5d.cwl | 120 +++++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 124 +++++++++--------- 4 files changed, 186 insertions(+), 66 deletions(-) create mode 100644 .renku/workflow/ed22b9067a714a60aa63af2ba94b8a5d.cwl diff --git a/.renku/workflow/ed22b9067a714a60aa63af2ba94b8a5d.cwl b/.renku/workflow/ed22b9067a714a60aa63af2ba94b8a5d.cwl new file mode 100644 index 000000000..e5710c831 --- /dev/null +++ b/.renku/workflow/ed22b9067a714a60aa63af2ba94b8a5d.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: states-daily.json + streamable: false + type: string + input_2: + default: states-metadata.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_1: + outputSource: step_3/output_1 + streamable: false + type: Directory + output_3: + outputSource: step_3/output_0 + streamable: false + type: File +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-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: + 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-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: [] + step_3: + in: + input_1: input_3 + input_2: input_4 + input_3: input_5 + input_4: input_6 + out: + - output_1 + - output_0 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 48da09c3c..2543d2dfd 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 sha256:6136c9676e800850762ab1152701a61382c92f61980841d108c11146a29bd9b1 -size 14530788 +oid sha256:89f20e92c5fb87d687ea74d5cdbb8e92e6a10c099199bbe4c21dc2b6d83ec9ff +size 14606134 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 271a33c1d..cf376b61d 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 sha256:99b0108722e9576f5d6a8d806819357999bb461cbc13f575db5c366bfbba325d -size 74414 +oid sha256:e220b9502811e769d29769330ed10ccda4981c9ee60ec1080bf538e82cd0f1e4 +size 75111 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 29dcbdf6d..f27824a6c 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": 3.758726, - "end_time": "2020-09-15T07:34:48.490913", + "duration": 3.099848, + "end_time": "2020-09-16T07:35:23.188379", "exception": false, - "start_time": "2020-09-15T07:34:44.732187", + "start_time": "2020-09-16T07:35:20.088531", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.029344, - "end_time": "2020-09-15T07:34:48.536522", + "duration": 0.029918, + "end_time": "2020-09-16T07:35:23.235828", "exception": false, - "start_time": "2020-09-15T07:34:48.507178", + "start_time": "2020-09-16T07:35:23.205910", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.033826, - "end_time": "2020-09-15T07:34:48.580975", + "duration": 0.026953, + "end_time": "2020-09-16T07:35:23.278213", "exception": false, - "start_time": "2020-09-15T07:34:48.547149", + "start_time": "2020-09-16T07:35:23.251260", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/jsxpi5jx/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/ahwdc6db/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": 0.010067, - "end_time": "2020-09-15T07:34:48.605607", + "duration": 0.008342, + "end_time": "2020-09-16T07:35:23.301372", "exception": false, - "start_time": "2020-09-15T07:34:48.595540", + "start_time": "2020-09-16T07:35:23.293030", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.82058, - "end_time": "2020-09-15T07:34:49.433425", + "duration": 0.6195, + "end_time": "2020-09-16T07:35:23.928157", "exception": false, - "start_time": "2020-09-15T07:34:48.612845", + "start_time": "2020-09-16T07:35:23.308657", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.027218, - "end_time": "2020-09-15T07:34:49.474403", + "duration": 0.0249, + "end_time": "2020-09-16T07:35:23.968133", "exception": false, - "start_time": "2020-09-15T07:34:49.447185", + "start_time": "2020-09-16T07:35:23.943233", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.094807, - "end_time": "2020-09-15T07:34:49.578347", + "duration": 0.081013, + "end_time": "2020-09-16T07:35:24.057991", "exception": false, - "start_time": "2020-09-15T07:34:49.483540", + "start_time": "2020-09-16T07:35:23.976978", "status": "completed" }, "tags": [] @@ -263,10 +263,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007654, - "end_time": "2020-09-15T07:34:49.599510", + "duration": 0.008085, + "end_time": "2020-09-16T07:35:24.081112", "exception": false, - "start_time": "2020-09-15T07:34:49.591856", + "start_time": "2020-09-16T07:35:24.073027", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 24.500853, - "end_time": "2020-09-15T07:35:14.107416", + "duration": 25.020287, + "end_time": "2020-09-16T07:35:49.108196", "exception": false, - "start_time": "2020-09-15T07:34:49.606563", + "start_time": "2020-09-16T07:35:24.087909", "status": "completed" }, "tags": [] @@ -300,10 +300,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.052058, - "end_time": "2020-09-15T07:35:14.175197", + "duration": 0.04623, + "end_time": "2020-09-16T07:35:49.168681", "exception": false, - "start_time": "2020-09-15T07:35:14.123139", + "start_time": "2020-09-16T07:35:49.122451", "status": "completed" }, "tags": [] @@ -322,10 +322,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.515353, - "end_time": "2020-09-15T07:35:14.706843", + "duration": 0.566828, + "end_time": "2020-09-16T07:35:49.749325", "exception": false, - "start_time": "2020-09-15T07:35:14.191490", + "start_time": "2020-09-16T07:35:49.182497", "status": "completed" }, "tags": [] @@ -335,7 +335,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "10906 data points\n" + "10962 data points\n" ] }, { @@ -385,21 +385,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200914</td>\n", + " <td>20200915</td>\n", " <td>AK</td>\n", - " <td>7270.0</td>\n", - " <td>396143.0</td>\n", + " <td>7313.0</td>\n", + " <td>399811.0</td>\n", " <td>NaN</td>\n", - " <td>37.0</td>\n", + " <td>44.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>8.0</td>\n", " <td>...</td>\n", - " <td>403413</td>\n", + " <td>407124</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>c47ba6fdd7b2d5856a69b861c773b64ba0dfcb76</td>\n", + " <td>b7d286bbeacaeef7cbaed188fba49cfe43c44450</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -409,21 +409,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200914</td>\n", + " <td>20200915</td>\n", " <td>AL</td>\n", - " <td>139459.0</td>\n", - " <td>888092.0</td>\n", + " <td>140160.0</td>\n", + " <td>892810.0</td>\n", " <td>NaN</td>\n", - " <td>792.0</td>\n", + " <td>716.0</td>\n", " <td>15756.0</td>\n", " <td>NaN</td>\n", - " <td>1611.0</td>\n", + " <td>1622.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>1027551</td>\n", - " <td>4</td>\n", - " <td>229</td>\n", - " <td>a852a907f21a1d2ec711d522217d2fa536b912e5</td>\n", + " <td>1032970</td>\n", + " <td>32</td>\n", + " <td>0</td>\n", + " <td>0f70ad393cc6c32f0b4ea5e8269035c8c655cc6d</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -438,20 +438,20 @@ ], "text/plain": [ " date state positive negative pending hospitalizedCurrently \\\n", - "0 20200914 AK 7270.0 396143.0 NaN 37.0 \n", - "1 20200914 AL 139459.0 888092.0 NaN 792.0 \n", + "0 20200915 AK 7313.0 399811.0 NaN 44.0 \n", + "1 20200915 AL 140160.0 892810.0 NaN 716.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 15756.0 NaN 1611.0 \n", + "1 15756.0 NaN 1622.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 8.0 ... 403413 0 0 \n", - "1 NaN ... 1027551 4 229 \n", + "0 8.0 ... 407124 0 0 \n", + "1 NaN ... 1032970 32 0 \n", "\n", " hash commercialScore \\\n", - "0 c47ba6fdd7b2d5856a69b861c773b64ba0dfcb76 0 \n", - "1 a852a907f21a1d2ec711d522217d2fa536b912e5 0 \n", + "0 b7d286bbeacaeef7cbaed188fba49cfe43c44450 0 \n", + "1 0f70ad393cc6c32f0b4ea5e8269035c8c655cc6d 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -491,18 +491,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 32.417203, - "end_time": "2020-09-15T07:35:15.138903", + "duration": 31.86503, + "end_time": "2020-09-16T07:35:50.171977", "environment_variables": {}, "exception": null, - "input_path": "/tmp/jsxpi5jx/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/ahwdc6db/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/jsxpi5jx/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/ahwdc6db/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-09-15T07:34:42.721700", + "start_time": "2020-09-16T07:35:18.306947", "version": "1.1.0" } }, -- GitLab