diff --git a/.renku/workflow/d6b47617f0be44b9b30d3dadf36ecbb5.cwl b/.renku/workflow/d6b47617f0be44b9b30d3dadf36ecbb5.cwl new file mode 100644 index 0000000000000000000000000000000000000000..f5144208b36236c200738058d75107082967a37a --- /dev/null +++ b/.renku/workflow/d6b47617f0be44b9b30d3dadf36ecbb5.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_2: + 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_0 + - output_1 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 313b4dcd8ba026393bf8d856e6f3ed39fb0fbff2..79108d0ddc56b64fd90e76fe08bc702e9649c7cc 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 sha256:3e3f573dfadd19add61f23f173c691376a9e7314f675c8c95d4d6587250003de -size 20320781 +oid sha256:bdad962f55521d67f534b2aeb1badaf6b703e76057653c269428a9e327f46b25 +size 20485516 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 338e4649aefc02038004ea48c352bed36bd9c059..3ae8e969bb56dd18b12fb5df9fea5fbbe6b8694b 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:ae36642b16f7dc2df25eba3634adfd5772731de2874d7bd625fe1b87d8a8bed7 -size 129621 +oid sha256:1d3ae035cdae441dea198f8a30c79af627e67e76952a06595f266a56428c44cc +size 132306 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 5e9efabf370f5c19464dfb67b3ed22b72c8639fb..b71ede25110f09d17878b684980a6aac9770fc2f 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": 5.255276, - "end_time": "2020-11-24T07:44:49.459961", + "duration": 6.475913, + "end_time": "2020-11-26T07:58:35.675162", "exception": false, - "start_time": "2020-11-24T07:44:44.204685", + "start_time": "2020-11-26T07:58:29.199249", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.023889, - "end_time": "2020-11-24T07:44:49.499512", + "duration": 0.020822, + "end_time": "2020-11-26T07:58:35.708619", "exception": false, - "start_time": "2020-11-24T07:44:49.475623", + "start_time": "2020-11-26T07:58:35.687797", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.010921, - "end_time": "2020-11-24T07:44:49.516597", + "duration": 0.017946, + "end_time": "2020-11-26T07:58:35.734541", "exception": false, - "start_time": "2020-11-24T07:44:49.505676", + "start_time": "2020-11-26T07:58:35.716595", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/t0y_essg/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/p4r5vl68/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": 0.004343, - "end_time": "2020-11-24T07:44:49.525362", + "duration": 0.005967, + "end_time": "2020-11-26T07:58:35.746416", "exception": false, - "start_time": "2020-11-24T07:44:49.521019", + "start_time": "2020-11-26T07:58:35.740449", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 10.391367, - "end_time": "2020-11-24T07:44:59.920943", + "duration": 10.821161, + "end_time": "2020-11-26T07:58:46.572554", "exception": false, - "start_time": "2020-11-24T07:44:49.529576", + "start_time": "2020-11-26T07:58:35.751393", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.024074, - "end_time": "2020-11-24T07:44:59.957871", + "duration": 0.029661, + "end_time": "2020-11-26T07:58:46.614568", "exception": false, - "start_time": "2020-11-24T07:44:59.933797", + "start_time": "2020-11-26T07:58:46.584907", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.066306, - "end_time": "2020-11-24T07:45:00.032836", + "duration": 0.063953, + "end_time": "2020-11-26T07:58:46.690388", "exception": false, - "start_time": "2020-11-24T07:44:59.966530", + "start_time": "2020-11-26T07:58:46.626435", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007437, - "end_time": "2020-11-24T07:45:00.052559", + "duration": 0.00631, + "end_time": "2020-11-26T07:58:46.707816", "exception": false, - "start_time": "2020-11-24T07:45:00.045122", + "start_time": "2020-11-26T07:58:46.701506", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 10.634454, - "end_time": "2020-11-24T07:45:10.693263", + "duration": 10.615746, + "end_time": "2020-11-26T07:58:57.330443", "exception": false, - "start_time": "2020-11-24T07:45:00.058809", + "start_time": "2020-11-26T07:58:46.714697", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.047788, - "end_time": "2020-11-24T07:45:10.754629", + "duration": 0.035601, + "end_time": "2020-11-26T07:58:57.378913", "exception": false, - "start_time": "2020-11-24T07:45:10.706841", + "start_time": "2020-11-26T07:58:57.343312", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.55723, - "end_time": "2020-11-24T07:45:11.324807", + "duration": 0.515161, + "end_time": "2020-11-26T07:58:57.904857", "exception": false, - "start_time": "2020-11-24T07:45:10.767577", + "start_time": "2020-11-26T07:58:57.389696", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "14899 data points\n" + "15017 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201123</td>\n", + " <td>20201125</td>\n", " <td>AK</td>\n", - " <td>27085.0</td>\n", + " <td>28339.0</td>\n", " <td>NaN</td>\n", - " <td>905931.0</td>\n", + " <td>925937.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>933016.0</td>\n", - " <td>146.0</td>\n", - " <td>617.0</td>\n", + " <td>954276.0</td>\n", + " <td>145.0</td>\n", + " <td>663.0</td>\n", " <td>...</td>\n", - " <td>933016</td>\n", + " <td>954276</td>\n", " <td>0</td>\n", - " <td>5</td>\n", - " <td>cbbe56430899e87f4ea8a0336bdc01fdb3b34998</td>\n", + " <td>32</td>\n", + " <td>e4015d41f79fe8f9552517702ee27996c0d86b6a</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>20201123</td>\n", + " <td>20201125</td>\n", " <td>AL</td>\n", - " <td>234080.0</td>\n", - " <td>38193.0</td>\n", - " <td>1337741.0</td>\n", + " <td>239318.0</td>\n", + " <td>39692.0</td>\n", + " <td>1353961.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1533628.0</td>\n", - " <td>1427.0</td>\n", - " <td>23942.0</td>\n", + " <td>1553587.0</td>\n", + " <td>1483.0</td>\n", + " <td>24376.0</td>\n", " <td>...</td>\n", - " <td>1571821</td>\n", - " <td>2</td>\n", - " <td>493</td>\n", - " <td>064b957eb226a47780acf699e8e184c49ced247b</td>\n", + " <td>1593279</td>\n", + " <td>60</td>\n", + " <td>216</td>\n", + " <td>cfbf3932514b39bd3f684d0fc235d8daa8009616</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 20201123 AK 27085.0 NaN 905931.0 NaN \n", - "1 20201123 AL 234080.0 38193.0 1337741.0 NaN \n", + "0 20201125 AK 28339.0 NaN 925937.0 NaN \n", + "1 20201125 AL 239318.0 39692.0 1353961.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 933016.0 146.0 \n", - "1 totalTestsViral 1533628.0 1427.0 \n", + "0 totalTestsViral 954276.0 145.0 \n", + "1 totalTestsViral 1553587.0 1483.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 617.0 ... 933016 0 5 \n", - "1 23942.0 ... 1571821 2 493 \n", + "0 663.0 ... 954276 0 32 \n", + "1 24376.0 ... 1593279 60 216 \n", "\n", " hash commercialScore \\\n", - "0 cbbe56430899e87f4ea8a0336bdc01fdb3b34998 0 \n", - "1 064b957eb226a47780acf699e8e184c49ced247b 0 \n", + "0 e4015d41f79fe8f9552517702ee27996c0d86b6a 0 \n", + "1 cfbf3932514b39bd3f684d0fc235d8daa8009616 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": 29.443362, - "end_time": "2020-11-24T07:45:11.646275", + "duration": 31.109234, + "end_time": "2020-11-26T07:58:58.230145", "environment_variables": {}, "exception": null, - "input_path": "/tmp/t0y_essg/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/p4r5vl68/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/t0y_essg/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/p4r5vl68/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-24T07:44:42.202913", + "start_time": "2020-11-26T07:58:27.120911", "version": "1.1.0" } },