From 42af95f6ded2cbab43d3d6ad51e44351ea339954 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Mon, 2 Nov 2020 07:40:05 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../975694ae2995400c97fb3932cbfc6bc1.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/975694ae2995400c97fb3932cbfc6bc1.cwl diff --git a/.renku/workflow/975694ae2995400c97fb3932cbfc6bc1.cwl b/.renku/workflow/975694ae2995400c97fb3932cbfc6bc1.cwl new file mode 100644 index 000000000..ee066a442 --- /dev/null +++ b/.renku/workflow/975694ae2995400c97fb3932cbfc6bc1.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: 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-metadata.json + streamable: false + type: string +outputs: + output_2: + 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-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: + input_1: input_2 + input_2: input_3 + input_3: input_4 + input_4: input_5 + out: + - output_1 + - output_0 + 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-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: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index a6c08f854..ed95224be 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 sha256:62d83a9e7e219b75a90a189b1afb7bc9b6bf64b0c1640f0a71d25ac2395c0017 -size 18536189 +oid sha256:136defefdadf4133e1d20028db3d077e4ea2d141eee730b645374fe90e0e94d7 +size 18613184 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index e377d225b..ab83d8c6d 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:e5c9982f4d81279d78cee5958a09ef5197dde0a9ac1d96cc2cb456bf4629da9d -size 108379 +oid sha256:f4ec8171dd99613eb1b81ecffa35355fe64aa13a460cc8b1a0dd3949e8933a1d +size 108213 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 014eaa71d..f8622d4e8 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": 5.037062, - "end_time": "2020-11-01T07:37:45.933708", + "duration": 4.38731, + "end_time": "2020-11-02T07:40:01.538008", "exception": false, - "start_time": "2020-11-01T07:37:40.896646", + "start_time": "2020-11-02T07:39:57.150698", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.02828, - "end_time": "2020-11-01T07:37:45.978315", + "duration": 0.024656, + "end_time": "2020-11-02T07:40:01.578884", "exception": false, - "start_time": "2020-11-01T07:37:45.950035", + "start_time": "2020-11-02T07:40:01.554228", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.022512, - "end_time": "2020-11-01T07:37:46.013706", + "duration": 0.027968, + "end_time": "2020-11-02T07:40:01.620053", "exception": false, - "start_time": "2020-11-01T07:37:45.991194", + "start_time": "2020-11-02T07:40:01.592085", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/y_sxckx7/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/qj7q4luy/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": 0.006401, - "end_time": "2020-11-01T07:37:46.029106", + "duration": 0.008125, + "end_time": "2020-11-02T07:40:01.638870", "exception": false, - "start_time": "2020-11-01T07:37:46.022705", + "start_time": "2020-11-02T07:40:01.630745", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.65798, - "end_time": "2020-11-01T07:37:46.693110", + "duration": 1.010529, + "end_time": "2020-11-02T07:40:02.656587", "exception": false, - "start_time": "2020-11-01T07:37:46.035130", + "start_time": "2020-11-02T07:40:01.646058", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.028713, - "end_time": "2020-11-01T07:37:46.735011", + "duration": 0.025484, + "end_time": "2020-11-02T07:40:02.696195", "exception": false, - "start_time": "2020-11-01T07:37:46.706298", + "start_time": "2020-11-02T07:40:02.670711", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.090248, - "end_time": "2020-11-01T07:37:46.838574", + "duration": 0.081021, + "end_time": "2020-11-02T07:40:02.785478", "exception": false, - "start_time": "2020-11-01T07:37:46.748326", + "start_time": "2020-11-02T07:40:02.704457", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007958, - "end_time": "2020-11-01T07:37:46.859707", + "duration": 0.008282, + "end_time": "2020-11-02T07:40:02.808385", "exception": false, - "start_time": "2020-11-01T07:37:46.851749", + "start_time": "2020-11-02T07:40:02.800103", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.437219, - "end_time": "2020-11-01T07:37:47.303780", + "duration": 0.486308, + "end_time": "2020-11-02T07:40:03.301771", "exception": false, - "start_time": "2020-11-01T07:37:46.866561", + "start_time": "2020-11-02T07:40:02.815463", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.063124, - "end_time": "2020-11-01T07:37:47.378855", + "duration": 0.06364, + "end_time": "2020-11-02T07:40:03.382133", "exception": false, - "start_time": "2020-11-01T07:37:47.315731", + "start_time": "2020-11-02T07:40:03.318493", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.650152, - "end_time": "2020-11-01T07:37:48.043379", + "duration": 0.707388, + "end_time": "2020-11-02T07:40:04.104186", "exception": false, - "start_time": "2020-11-01T07:37:47.393227", + "start_time": "2020-11-02T07:40:03.396798", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13607 data points\n" + "13663 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201031</td>\n", + " <td>20201101</td>\n", " <td>AK</td>\n", - " <td>16355.0</td>\n", + " <td>16705.0</td>\n", " <td>NaN</td>\n", " <td>587852.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", " <td>604207.0</td>\n", - " <td>94.0</td>\n", + " <td>98.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>604207</td>\n", + " <td>604557</td>\n", " <td>1</td>\n", " <td>0</td>\n", - " <td>4fd1e9352fd403f75cd0f4b8c7d420de8eb73f0b</td>\n", + " <td>0fe81cb4b3a17588cfde8bbd309c3f0f2e786fd2</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -426,21 +426,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20201031</td>\n", + " <td>20201101</td>\n", " <td>AL</td>\n", - " <td>192285.0</td>\n", - " <td>27990.0</td>\n", - " <td>1181958.0</td>\n", + " <td>193985.0</td>\n", + " <td>28746.0</td>\n", + " <td>1187341.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1346253.0</td>\n", - " <td>960.0</td>\n", + " <td>1352580.0</td>\n", + " <td>967.0</td>\n", " <td>20450.0</td>\n", " <td>...</td>\n", - " <td>1374243</td>\n", - " <td>35</td>\n", + " <td>1381326</td>\n", + " <td>6</td>\n", " <td>0</td>\n", - " <td>5194e1ea04ad4dace9ab68daca52f9fa98e39f1b</td>\n", + " <td>0bd0f2572848cf6f46a651c076d32afceb8dc6a1</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -455,20 +455,20 @@ ], "text/plain": [ " date state positive probableCases negative pending \\\n", - "0 20201031 AK 16355.0 NaN 587852.0 NaN \n", - "1 20201031 AL 192285.0 27990.0 1181958.0 NaN \n", + "0 20201101 AK 16705.0 NaN 587852.0 NaN \n", + "1 20201101 AL 193985.0 28746.0 1187341.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 604207.0 94.0 \n", - "1 totalTestsViral 1346253.0 960.0 \n", + "0 totalTestsViral 604207.0 98.0 \n", + "1 totalTestsViral 1352580.0 967.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 604207 1 0 \n", - "1 20450.0 ... 1374243 35 0 \n", + "0 NaN ... 604557 1 0 \n", + "1 20450.0 ... 1381326 6 0 \n", "\n", " hash commercialScore \\\n", - "0 4fd1e9352fd403f75cd0f4b8c7d420de8eb73f0b 0 \n", - "1 5194e1ea04ad4dace9ab68daca52f9fa98e39f1b 0 \n", + "0 0fe81cb4b3a17588cfde8bbd309c3f0f2e786fd2 0 \n", + "1 0bd0f2572848cf6f46a651c076d32afceb8dc6a1 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -508,18 +508,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 9.337475, - "end_time": "2020-11-01T07:37:48.472569", + "duration": 9.174658, + "end_time": "2020-11-02T07:40:04.530883", "environment_variables": {}, "exception": null, - "input_path": "/tmp/y_sxckx7/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/qj7q4luy/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/y_sxckx7/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/qj7q4luy/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-01T07:37:39.135094", + "start_time": "2020-11-02T07:39:55.356225", "version": "1.1.0" } }, -- GitLab