From ac17e602f599e7fe55c746e01990d09fcdfcefb4 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Sat, 12 Dec 2020 07:38:08 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../3f85941e1428451fb4b69ba4d17a2f68.cwl | 120 ++++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 136 +++++++++--------- 4 files changed, 192 insertions(+), 72 deletions(-) create mode 100644 .renku/workflow/3f85941e1428451fb4b69ba4d17a2f68.cwl diff --git a/.renku/workflow/3f85941e1428451fb4b69ba4d17a2f68.cwl b/.renku/workflow/3f85941e1428451fb4b69ba4d17a2f68.cwl new file mode 100644 index 000000000..308b84ab3 --- /dev/null +++ b/.renku/workflow/3f85941e1428451fb4b69ba4d17a2f68.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_0: + outputSource: step_1/output_1 + streamable: false + type: Directory + output_2: + 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 3c2976451..b155be792 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:db2f6e5f2d8c29067e43dc70ea18465841f0a1851b624b085e4d8a891a81bf20 -size 21571126 +oid sha256:e62d080a170bc871e31539c8be8b3cdd3658cbe5d7fcf8a589eee3aefd0209e4 +size 21724952 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index bfe4fe912..4a863b13d 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:d9833f5095ca0a847f00ab5622ab1be1ce68bcf61e3005431bb99627cdcd43ec -size 149725 +oid sha256:f2f507c8365cc03aedb8456a10e41bdea80f74fbd6ed14b16746c6968065691b +size 150268 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 08cf9764e..ba4c0680b 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": 6.57455, - "end_time": "2020-12-10T07:45:03.853676", + "duration": 6.889652, + "end_time": "2020-12-12T07:38:05.389643", "exception": false, - "start_time": "2020-12-10T07:44:57.279126", + "start_time": "2020-12-12T07:37:58.499991", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.023212, - "end_time": "2020-12-10T07:45:03.894419", + "duration": 0.015644, + "end_time": "2020-12-12T07:38:05.417406", "exception": false, - "start_time": "2020-12-10T07:45:03.871207", + "start_time": "2020-12-12T07:38:05.401762", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.026984, - "end_time": "2020-12-10T07:45:03.932218", + "duration": 0.010775, + "end_time": "2020-12-12T07:38:05.433035", "exception": false, - "start_time": "2020-12-10T07:45:03.905234", + "start_time": "2020-12-12T07:38:05.422260", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/hjio672c/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/r6lieysd/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": 0.009132, - "end_time": "2020-12-10T07:45:03.952183", + "duration": 0.004017, + "end_time": "2020-12-12T07:38:05.441138", "exception": false, - "start_time": "2020-12-10T07:45:03.943051", + "start_time": "2020-12-12T07:38:05.437121", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.838065, - "end_time": "2020-12-10T07:45:04.800589", + "duration": 0.343477, + "end_time": "2020-12-12T07:38:05.788838", "exception": false, - "start_time": "2020-12-10T07:45:03.962524", + "start_time": "2020-12-12T07:38:05.445361", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.031905, - "end_time": "2020-12-10T07:45:04.847192", + "duration": 0.022185, + "end_time": "2020-12-12T07:38:05.822402", "exception": false, - "start_time": "2020-12-10T07:45:04.815287", + "start_time": "2020-12-12T07:38:05.800217", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.114974, - "end_time": "2020-12-10T07:45:04.976342", + "duration": 0.071164, + "end_time": "2020-12-12T07:38:05.902998", "exception": false, - "start_time": "2020-12-10T07:45:04.861368", + "start_time": "2020-12-12T07:38:05.831834", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.011016, - "end_time": "2020-12-10T07:45:05.004437", + "duration": 0.006768, + "end_time": "2020-12-12T07:38:05.921404", "exception": false, - "start_time": "2020-12-10T07:45:04.993421", + "start_time": "2020-12-12T07:38:05.914636", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.641805, - "end_time": "2020-12-10T07:45:05.655807", + "duration": 0.644425, + "end_time": "2020-12-12T07:38:06.571920", "exception": false, - "start_time": "2020-12-10T07:45:05.014002", + "start_time": "2020-12-12T07:38:05.927495", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.055563, - "end_time": "2020-12-10T07:45:05.724800", + "duration": 0.050735, + "end_time": "2020-12-12T07:38:06.639504", "exception": false, - "start_time": "2020-12-10T07:45:05.669237", + "start_time": "2020-12-12T07:38:06.588769", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.711583, - "end_time": "2020-12-10T07:45:06.452035", + "duration": 0.614728, + "end_time": "2020-12-12T07:38:07.266668", "exception": false, - "start_time": "2020-12-10T07:45:05.740452", + "start_time": "2020-12-12T07:38:06.651940", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "15801 data points\n" + "15913 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201209</td>\n", + " <td>20201211</td>\n", " <td>AK</td>\n", - " <td>37342.0</td>\n", + " <td>38584.0</td>\n", " <td>NaN</td>\n", - " <td>1062213.0</td>\n", + " <td>1084961.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1099555.0</td>\n", - " <td>165.0</td>\n", - " <td>847.0</td>\n", + " <td>1123545.0</td>\n", + " <td>146.0</td>\n", + " <td>869.0</td>\n", " <td>...</td>\n", - " <td>1099555</td>\n", - " <td>4</td>\n", - " <td>20</td>\n", - " <td>1ff1db7b83dd9155a28a2e3687fbbb65d619643e</td>\n", + " <td>1123545</td>\n", + " <td>3</td>\n", + " <td>9</td>\n", + " <td>47f2cb8f9f01019549b85b0450c8c42c15bb3e77</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>20201209</td>\n", + " <td>20201211</td>\n", " <td>AL</td>\n", - " <td>280187.0</td>\n", - " <td>48648.0</td>\n", - " <td>1441625.0</td>\n", + " <td>288775.0</td>\n", + " <td>50896.0</td>\n", + " <td>1458560.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsPeopleViral</td>\n", - " <td>1673164.0</td>\n", + " <td>1696439.0</td>\n", " <td>2111.0</td>\n", - " <td>27618.0</td>\n", + " <td>28146.0</td>\n", " <td>...</td>\n", - " <td>1721812</td>\n", - " <td>43</td>\n", - " <td>295</td>\n", - " <td>e16ff21c63db24182a3474993772040770f37424</td>\n", + " <td>1747335</td>\n", + " <td>52</td>\n", + " <td>528</td>\n", + " <td>3bf1ccc167c9d139e7506942ea8f53c42bdeeab7</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 20201209 AK 37342.0 NaN 1062213.0 NaN \n", - "1 20201209 AL 280187.0 48648.0 1441625.0 NaN \n", + "0 20201211 AK 38584.0 NaN 1084961.0 NaN \n", + "1 20201211 AL 288775.0 50896.0 1458560.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 1099555.0 165.0 \n", - "1 totalTestsPeopleViral 1673164.0 2111.0 \n", + "0 totalTestsViral 1123545.0 146.0 \n", + "1 totalTestsPeopleViral 1696439.0 2111.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 847.0 ... 1099555 4 20 \n", - "1 27618.0 ... 1721812 43 295 \n", + "0 869.0 ... 1123545 3 9 \n", + "1 28146.0 ... 1747335 52 528 \n", "\n", " hash commercialScore \\\n", - "0 1ff1db7b83dd9155a28a2e3687fbbb65d619643e 0 \n", - "1 e16ff21c63db24182a3474993772040770f37424 0 \n", + "0 47f2cb8f9f01019549b85b0450c8c42c15bb3e77 0 \n", + "1 3bf1ccc167c9d139e7506942ea8f53c42bdeeab7 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": 13.022997, - "end_time": "2020-12-10T07:45:06.880109", + "duration": 10.7272, + "end_time": "2020-12-12T07:38:07.588984", "environment_variables": {}, "exception": null, - "input_path": "/tmp/hjio672c/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/r6lieysd/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/hjio672c/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/r6lieysd/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-12-10T07:44:53.857112", + "start_time": "2020-12-12T07:37:56.861784", "version": "1.1.0" } }, -- GitLab