From aab343e86dba51a6339c2ff72d4c42149e4965c0 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Thu, 6 Aug 2020 07:32:31 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../e9dce04f985d403582259868f7c5f7cb.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/e9dce04f985d403582259868f7c5f7cb.cwl diff --git a/.renku/workflow/e9dce04f985d403582259868f7c5f7cb.cwl b/.renku/workflow/e9dce04f985d403582259868f7c5f7cb.cwl new file mode 100644 index 000000000..855541b73 --- /dev/null +++ b/.renku/workflow/e9dce04f985d403582259868f7c5f7cb.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_1: + 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 b651e7865..4e3362939 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 sha256:3c176e31513d08dbf8d4e7af65cd1eb8a248b1c36c344bae39e7e2dd9a3e7410 -size 7922705 +oid sha256:157ea87a5e7a61335b9fc48a50577ac43ef0bce919622996f0e51176e5d1e01a +size 7975364 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 4dfd520f5..c96a14c3f 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 sha256:260a3d2511a33a3c73ad023e9856bcc6ea3276e38025017272db886c52e0a5be -size 57814 +oid sha256:ebba936994c56fd0e6e971917ebe1e08e5d48143d430ce6e2aa4ec2d2e0972f0 +size 58914 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 572fb57d6..33071a060 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": 3.880489, - "end_time": "2020-08-05T07:33:37.928356", + "duration": 4.25667, + "end_time": "2020-08-06T07:32:29.957512", "exception": false, - "start_time": "2020-08-05T07:33:34.047867", + "start_time": "2020-08-06T07:32:25.700842", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.024466, - "end_time": "2020-08-05T07:33:37.961210", + "duration": 0.020099, + "end_time": "2020-08-06T07:32:29.990655", "exception": false, - "start_time": "2020-08-05T07:33:37.936744", + "start_time": "2020-08-06T07:32:29.970556", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.024783, - "end_time": "2020-08-05T07:33:37.993930", + "duration": 0.019877, + "end_time": "2020-08-06T07:32:30.019779", "exception": false, - "start_time": "2020-08-05T07:33:37.969147", + "start_time": "2020-08-06T07:32:29.999902", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/ooj1q6z_/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/y8yxp7e2/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": 0.007393, - "end_time": "2020-08-05T07:33:38.011199", + "duration": 0.007191, + "end_time": "2020-08-06T07:32:30.035387", "exception": false, - "start_time": "2020-08-05T07:33:38.003806", + "start_time": "2020-08-06T07:32:30.028196", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.174305, - "end_time": "2020-08-05T07:33:38.191736", + "duration": 0.067951, + "end_time": "2020-08-06T07:32:30.110740", "exception": false, - "start_time": "2020-08-05T07:33:38.017431", + "start_time": "2020-08-06T07:32:30.042789", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.030909, - "end_time": "2020-08-05T07:33:38.238139", + "duration": 0.022383, + "end_time": "2020-08-06T07:32:30.144709", "exception": false, - "start_time": "2020-08-05T07:33:38.207230", + "start_time": "2020-08-06T07:32:30.122326", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.081414, - "end_time": "2020-08-05T07:33:38.332447", + "duration": 0.09242, + "end_time": "2020-08-06T07:32:30.245483", "exception": false, - "start_time": "2020-08-05T07:33:38.251033", + "start_time": "2020-08-06T07:32:30.153063", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008205, - "end_time": "2020-08-05T07:33:38.352814", + "duration": 0.008447, + "end_time": "2020-08-06T07:32:30.264032", "exception": false, - "start_time": "2020-08-05T07:33:38.344609", + "start_time": "2020-08-06T07:32:30.255585", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.242617, - "end_time": "2020-08-05T07:33:38.602651", + "duration": 0.272467, + "end_time": "2020-08-06T07:32:30.545373", "exception": false, - "start_time": "2020-08-05T07:33:38.360034", + "start_time": "2020-08-06T07:32:30.272906", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.037749, - "end_time": "2020-08-05T07:33:38.658314", + "duration": 0.03551, + "end_time": "2020-08-06T07:32:30.593640", "exception": false, - "start_time": "2020-08-05T07:33:38.620565", + "start_time": "2020-08-06T07:32:30.558130", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.326858, - "end_time": "2020-08-05T07:33:38.997546", + "duration": 0.339374, + "end_time": "2020-08-06T07:32:30.944934", "exception": false, - "start_time": "2020-08-05T07:33:38.670688", + "start_time": "2020-08-06T07:32:30.605560", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "8529 data points\n" + "8585 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200804</td>\n", + " <td>20200805</td>\n", " <td>AK</td>\n", - " <td>4125.0</td>\n", - " <td>248524.0</td>\n", + " <td>4183.0</td>\n", + " <td>253096.0</td>\n", " <td>NaN</td>\n", - " <td>33.0</td>\n", + " <td>38.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>2.0</td>\n", " <td>...</td>\n", - " <td>252649</td>\n", + " <td>257279</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>59a8cabc3e35a5d51323bff0c43feacfd0806dc2</td>\n", + " <td>0a85599b284b6519c9ef59c495ee7e02bb9160a5</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>20200804</td>\n", + " <td>20200805</td>\n", " <td>AL</td>\n", - " <td>93702.0</td>\n", - " <td>628599.0</td>\n", + " <td>94654.0</td>\n", + " <td>632806.0</td>\n", " <td>NaN</td>\n", - " <td>1517.0</td>\n", + " <td>1575.0</td>\n", " <td>11100.0</td>\n", " <td>NaN</td>\n", - " <td>1173.0</td>\n", + " <td>1200.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>722301</td>\n", - " <td>33</td>\n", - " <td>235</td>\n", - " <td>000ce6dde752f7fbcd8a4262d549da464dfabb5c</td>\n", + " <td>727460</td>\n", + " <td>29</td>\n", + " <td>0</td>\n", + " <td>672da264092f08c418d1945dcb919436e9d82d2f</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 20200804 AK 4125.0 248524.0 NaN 33.0 \n", - "1 20200804 AL 93702.0 628599.0 NaN 1517.0 \n", + "0 20200805 AK 4183.0 253096.0 NaN 38.0 \n", + "1 20200805 AL 94654.0 632806.0 NaN 1575.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 11100.0 NaN 1173.0 \n", + "1 11100.0 NaN 1200.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 2.0 ... 252649 0 0 \n", - "1 NaN ... 722301 33 235 \n", + "0 2.0 ... 257279 0 0 \n", + "1 NaN ... 727460 29 0 \n", "\n", " hash commercialScore \\\n", - "0 59a8cabc3e35a5d51323bff0c43feacfd0806dc2 0 \n", - "1 000ce6dde752f7fbcd8a4262d549da464dfabb5c 0 \n", + "0 0a85599b284b6519c9ef59c495ee7e02bb9160a5 0 \n", + "1 672da264092f08c418d1945dcb919436e9d82d2f 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": 6.958997, - "end_time": "2020-08-05T07:33:39.426894", + "duration": 7.007186, + "end_time": "2020-08-06T07:32:31.268636", "environment_variables": {}, "exception": null, - "input_path": "/tmp/ooj1q6z_/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/y8yxp7e2/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/ooj1q6z_/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/y8yxp7e2/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-08-05T07:33:32.467897", + "start_time": "2020-08-06T07:32:24.261450", "version": "1.1.0" } }, -- GitLab