From 925135e1ea99555cf595046ca934059444146d19 Mon Sep 17 00:00:00 2001 From: "CR (covid cron)" <beepbop@example.com> Date: Mon, 7 Dec 2020 07:37:58 +0000 Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json data/covidtracking/states-daily.json --- .../d960da088bd94329ae5215b9d22ad8d8.cwl | 120 ++++++++++++++++ data/covidtracking/states-daily.json | 4 +- data/covidtracking/states-metadata.json | 4 +- runs/download-covidtracking-data.runs.ipynb | 132 +++++++++--------- 4 files changed, 190 insertions(+), 70 deletions(-) create mode 100644 .renku/workflow/d960da088bd94329ae5215b9d22ad8d8.cwl diff --git a/.renku/workflow/d960da088bd94329ae5215b9d22ad8d8.cwl b/.renku/workflow/d960da088bd94329ae5215b9d22ad8d8.cwl new file mode 100644 index 000000000..e4b2b04ae --- /dev/null +++ b/.renku/workflow/d960da088bd94329ae5215b9d22ad8d8.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-daily.json + streamable: false + type: string + input_6: + default: states-metadata.json + streamable: false + type: string +outputs: + output_2: + outputSource: step_1/output_0 + streamable: false + type: File + output_3: + outputSource: step_1/output_1 + streamable: false + type: Directory +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-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_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-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 87afbffbe..6166bccda 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:a3628828503580ed48823aa574369330b240f7b22a454de93d09f818461ce7ed -size 21258293 +oid sha256:1ca31123bba8b2cb17227b07d08312094fa1fdeb67294266da73e6fc4813e115 +size 21335523 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 1e43dae89..76b115006 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 sha256:8bcfdaa5409e4f9953474c5f60a032653b52d65518107a94d77b82bb2eb334ff -size 146254 +oid sha256:a53136aa9dcd2c2aa2979195f39d9f5c0e68742b4abdaad373c519dc2337545c +size 146464 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 2d191c18c..3bc594c00 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": 4.717714, - "end_time": "2020-12-06T07:36:42.848851", + "duration": 4.152565, + "end_time": "2020-12-07T07:37:54.515472", "exception": false, - "start_time": "2020-12-06T07:36:38.131137", + "start_time": "2020-12-07T07:37:50.362907", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.020975, - "end_time": "2020-12-06T07:36:42.884296", + "duration": 0.020423, + "end_time": "2020-12-07T07:37:54.550390", "exception": false, - "start_time": "2020-12-06T07:36:42.863321", + "start_time": "2020-12-07T07:37:54.529967", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.018155, - "end_time": "2020-12-06T07:36:42.910185", + "duration": 0.018496, + "end_time": "2020-12-07T07:37:54.576647", "exception": false, - "start_time": "2020-12-06T07:36:42.892030", + "start_time": "2020-12-07T07:37:54.558151", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/bgj85y8_/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/x6h2bjc3/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": 0.005918, - "end_time": "2020-12-06T07:36:42.922347", + "duration": 0.006579, + "end_time": "2020-12-07T07:37:54.590047", "exception": false, - "start_time": "2020-12-06T07:36:42.916429", + "start_time": "2020-12-07T07:37:54.583468", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.836113, - "end_time": "2020-12-06T07:36:43.763079", + "duration": 0.69645, + "end_time": "2020-12-07T07:37:55.291679", "exception": false, - "start_time": "2020-12-06T07:36:42.926966", + "start_time": "2020-12-07T07:37:54.595229", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.022803, - "end_time": "2020-12-06T07:36:43.797693", + "duration": 0.022024, + "end_time": "2020-12-07T07:37:55.325785", "exception": false, - "start_time": "2020-12-06T07:36:43.774890", + "start_time": "2020-12-07T07:37:55.303761", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.104921, - "end_time": "2020-12-06T07:36:43.911415", + "duration": 0.069775, + "end_time": "2020-12-07T07:37:55.403880", "exception": false, - "start_time": "2020-12-06T07:36:43.806494", + "start_time": "2020-12-07T07:37:55.334105", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007727, - "end_time": "2020-12-06T07:36:43.931182", + "duration": 0.007851, + "end_time": "2020-12-07T07:37:55.425086", "exception": false, - "start_time": "2020-12-06T07:36:43.923455", + "start_time": "2020-12-07T07:37:55.417235", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.674744, - "end_time": "2020-12-06T07:36:44.611412", + "duration": 0.741031, + "end_time": "2020-12-07T07:37:56.172694", "exception": false, - "start_time": "2020-12-06T07:36:43.936668", + "start_time": "2020-12-07T07:37:55.431663", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.043869, - "end_time": "2020-12-06T07:36:44.670424", + "duration": 0.0364, + "end_time": "2020-12-07T07:37:56.223660", "exception": false, - "start_time": "2020-12-06T07:36:44.626555", + "start_time": "2020-12-07T07:37:56.187260", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.584751, - "end_time": "2020-12-06T07:36:45.268667", + "duration": 0.475217, + "end_time": "2020-12-07T07:37:56.707381", "exception": false, - "start_time": "2020-12-06T07:36:44.683916", + "start_time": "2020-12-07T07:37:56.232164", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "15577 data points\n" + "15633 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201205</td>\n", + " <td>20201206</td>\n", " <td>AK</td>\n", - " <td>34963.0</td>\n", + " <td>35720.0</td>\n", " <td>NaN</td>\n", - " <td>1032268.0</td>\n", + " <td>1042056.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1067231.0</td>\n", - " <td>166.0</td>\n", + " <td>1077776.0</td>\n", + " <td>164.0</td>\n", " <td>799.0</td>\n", " <td>...</td>\n", - " <td>1067231</td>\n", - " <td>1</td>\n", - " <td>5</td>\n", - " <td>d8b27967afb185d8e4f38f5209d713ef23065d88</td>\n", + " <td>1077776</td>\n", + " <td>0</td>\n", + " <td>0</td>\n", + " <td>7b1d31e2756687bb9259b29195f1db6cdb321ea6</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>20201205</td>\n", + " <td>20201206</td>\n", " <td>AL</td>\n", - " <td>267589.0</td>\n", - " <td>45785.0</td>\n", - " <td>1415357.0</td>\n", + " <td>269877.0</td>\n", + " <td>45962.0</td>\n", + " <td>1421126.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsPeopleViral</td>\n", - " <td>1637161.0</td>\n", - " <td>1867.0</td>\n", + " <td>1645041.0</td>\n", + " <td>1927.0</td>\n", " <td>26331.0</td>\n", " <td>...</td>\n", - " <td>1682946</td>\n", - " <td>46</td>\n", + " <td>1691003</td>\n", + " <td>12</td>\n", " <td>0</td>\n", - " <td>7a521812788c45e903b5a08f357bdd78f668340a</td>\n", + " <td>19454ed8fe28fc0a7948fc0771b2f3c846c1c92e</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 20201205 AK 34963.0 NaN 1032268.0 NaN \n", - "1 20201205 AL 267589.0 45785.0 1415357.0 NaN \n", + "0 20201206 AK 35720.0 NaN 1042056.0 NaN \n", + "1 20201206 AL 269877.0 45962.0 1421126.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 1067231.0 166.0 \n", - "1 totalTestsPeopleViral 1637161.0 1867.0 \n", + "0 totalTestsViral 1077776.0 164.0 \n", + "1 totalTestsPeopleViral 1645041.0 1927.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 799.0 ... 1067231 1 5 \n", - "1 26331.0 ... 1682946 46 0 \n", + "0 799.0 ... 1077776 0 0 \n", + "1 26331.0 ... 1691003 12 0 \n", "\n", " hash commercialScore \\\n", - "0 d8b27967afb185d8e4f38f5209d713ef23065d88 0 \n", - "1 7a521812788c45e903b5a08f357bdd78f668340a 0 \n", + "0 7b1d31e2756687bb9259b29195f1db6cdb321ea6 0 \n", + "1 19454ed8fe28fc0a7948fc0771b2f3c846c1c92e 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": 9.124953, - "end_time": "2020-12-06T07:36:45.590793", + "duration": 8.764442, + "end_time": "2020-12-07T07:37:57.030694", "environment_variables": {}, "exception": null, - "input_path": "/tmp/bgj85y8_/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/x6h2bjc3/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/bgj85y8_/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/x6h2bjc3/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-12-06T07:36:36.465840", + "start_time": "2020-12-07T07:37:48.266252", "version": "1.1.0" } }, -- GitLab