diff --git a/.renku/workflow/044798d41dd4416b8d5c9355e16d551a.cwl b/.renku/workflow/044798d41dd4416b8d5c9355e16d551a.cwl new file mode 100644 index 0000000000000000000000000000000000000000..fde0deca2576a4283f0bc4ad475ba0a2c8b46edd --- /dev/null +++ b/.renku/workflow/044798d41dd4416b8d5c9355e16d551a.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_0: + outputSource: step_3/output_0 + streamable: false + type: File + output_2: + outputSource: step_3/output_1 + streamable: false + type: Directory +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_1 + - output_0 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index f6d605a83690cde7d628276bc3b8c4915fe3b1fe..fbc87dfdb96042130c44350efc976e4b0c5782d8 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 sha256:56ac30fed3a75c442559f4d888dadb545890e7ac864db45a1eb466a31ad4090f -size 491877 +oid sha256:1b4cf0b6b4063633188335e2ac325c164482417d490642fd7bc2d28c999c0a16 +size 512308 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 464bdcf241c2e20b932f5ab998cf20d18b400c50..5ea208db896ce30ce126d6be413f55538bae9d2b 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:d7100b6f41ef1cfea9a6a995dc202bcb62aa8110ff6b5605c05e7ba1fd721794 -size 25296 +oid sha256:162a1e7cff5c197e79e0efcbfe69bd4c94d8872d0b7c143fe9286680489fa5a9 +size 25370 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 4f842f2090c7b93cd9893169c8a79ed031ab7316..0ca3b1c28caa3bb4b2a5d989a7431a3e229162d4 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": 0.471733, - "end_time": "2020-03-31T08:24:30.934689", + "duration": 0.486177, + "end_time": "2020-04-01T11:24:57.200104", "exception": false, - "start_time": "2020-03-31T08:24:30.462956", + "start_time": "2020-04-01T11:24:56.713927", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.026018, - "end_time": "2020-03-31T08:24:30.978110", + "duration": 0.019035, + "end_time": "2020-04-01T11:24:57.231060", "exception": false, - "start_time": "2020-03-31T08:24:30.952092", + "start_time": "2020-04-01T11:24:57.212025", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.023835, - "end_time": "2020-03-31T08:24:31.015903", + "duration": 0.023334, + "end_time": "2020-04-01T11:24:57.261409", "exception": false, - "start_time": "2020-03-31T08:24:30.992068", + "start_time": "2020-04-01T11:24:57.238075", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/anascddl/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/v_x6sgie/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": 0.005891, - "end_time": "2020-03-31T08:24:31.031893", + "duration": 0.007865, + "end_time": "2020-04-01T11:24:57.279101", "exception": false, - "start_time": "2020-03-31T08:24:31.026002", + "start_time": "2020-04-01T11:24:57.271236", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.748106, - "end_time": "2020-03-31T08:24:31.785977", + "duration": 0.195872, + "end_time": "2020-04-01T11:24:57.481715", "exception": false, - "start_time": "2020-03-31T08:24:31.037871", + "start_time": "2020-04-01T11:24:57.285843", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.025387, - "end_time": "2020-03-31T08:24:31.825517", + "duration": 0.024992, + "end_time": "2020-04-01T11:24:57.512245", "exception": false, - "start_time": "2020-03-31T08:24:31.800130", + "start_time": "2020-04-01T11:24:57.487253", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.072397, - "end_time": "2020-03-31T08:24:31.905584", + "duration": 0.0607, + "end_time": "2020-04-01T11:24:57.580990", "exception": false, - "start_time": "2020-03-31T08:24:31.833187", + "start_time": "2020-04-01T11:24:57.520290", "status": "completed" }, "tags": [] @@ -246,10 +246,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.00763, - "end_time": "2020-03-31T08:24:31.927988", + "duration": 0.007609, + "end_time": "2020-04-01T11:24:57.607059", "exception": false, - "start_time": "2020-03-31T08:24:31.920358", + "start_time": "2020-04-01T11:24:57.599450", "status": "completed" }, "tags": [] @@ -263,10 +263,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.330782, - "end_time": "2020-03-31T08:24:32.265799", + "duration": 0.199119, + "end_time": "2020-04-01T11:24:57.812781", "exception": false, - "start_time": "2020-03-31T08:24:31.935017", + "start_time": "2020-04-01T11:24:57.613662", "status": "completed" }, "tags": [] @@ -283,10 +283,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.029205, - "end_time": "2020-03-31T08:24:32.310572", + "duration": 0.027022, + "end_time": "2020-04-01T11:24:57.854471", "exception": false, - "start_time": "2020-03-31T08:24:32.281367", + "start_time": "2020-04-01T11:24:57.827449", "status": "completed" }, "tags": [] @@ -305,10 +305,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.07432, - "end_time": "2020-03-31T08:24:32.406024", + "duration": 0.082088, + "end_time": "2020-04-01T11:24:57.946250", "exception": false, - "start_time": "2020-03-31T08:24:32.331704", + "start_time": "2020-04-01T11:24:57.864162", "status": "completed" }, "tags": [] @@ -318,7 +318,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "1373 data points\n" + "1429 data points\n" ] }, { @@ -364,43 +364,43 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200330</td>\n", + " <td>20200331</td>\n", " <td>AK</td>\n", - " <td>114.0</td>\n", - " <td>3540.0</td>\n", + " <td>119.0</td>\n", + " <td>3594.0</td>\n", " <td>NaN</td>\n", " <td>7.0</td>\n", " <td>3.0</td>\n", - " <td>3654</td>\n", - " <td>01a1c96fd2ed214d8747ab778c2fec7203c8cd2f</td>\n", - " <td>2020-03-30T20:00:00Z</td>\n", - " <td>3654</td>\n", + " <td>3713</td>\n", + " <td>5339d73f78797a2174fe32109c081c80915f8378</td>\n", + " <td>2020-03-31T20:00:00Z</td>\n", + " <td>3713</td>\n", " <td>2</td>\n", - " <td>1.0</td>\n", - " <td>1.0</td>\n", - " <td>308.0</td>\n", - " <td>12.0</td>\n", - " <td>320.0</td>\n", + " <td>0.0</td>\n", + " <td>0.0</td>\n", + " <td>54.0</td>\n", + " <td>5.0</td>\n", + " <td>59.0</td>\n", " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200330</td>\n", + " <td>20200331</td>\n", " <td>AL</td>\n", - " <td>859.0</td>\n", - " <td>5694.0</td>\n", + " <td>981.0</td>\n", + " <td>6298.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", - " <td>6.0</td>\n", - " <td>6553</td>\n", - " <td>1ced1dbd9879f8bbc4b1f7b7876b82611895d58e</td>\n", - " <td>2020-03-30T20:00:00Z</td>\n", - " <td>6553</td>\n", + " <td>13.0</td>\n", + " <td>7279</td>\n", + " <td>06fc31f03af09c0f09f4c5bc18fac9daa63e94f8</td>\n", + " <td>2020-03-31T20:00:00Z</td>\n", + " <td>7279</td>\n", " <td>1</td>\n", - " <td>2.0</td>\n", + " <td>7.0</td>\n", " <td>0.0</td>\n", - " <td>1510.0</td>\n", - " <td>53.0</td>\n", - " <td>1563.0</td>\n", + " <td>604.0</td>\n", + " <td>122.0</td>\n", + " <td>726.0</td>\n", " </tr>\n", " </tbody>\n", "</table>\n", @@ -408,20 +408,20 @@ ], "text/plain": [ " date state positive negative pending hospitalized death total \\\n", - "0 20200330 AK 114.0 3540.0 NaN 7.0 3.0 3654 \n", - "1 20200330 AL 859.0 5694.0 NaN NaN 6.0 6553 \n", + "0 20200331 AK 119.0 3594.0 NaN 7.0 3.0 3713 \n", + "1 20200331 AL 981.0 6298.0 NaN NaN 13.0 7279 \n", "\n", " hash dateChecked \\\n", - "0 01a1c96fd2ed214d8747ab778c2fec7203c8cd2f 2020-03-30T20:00:00Z \n", - "1 1ced1dbd9879f8bbc4b1f7b7876b82611895d58e 2020-03-30T20:00:00Z \n", + "0 5339d73f78797a2174fe32109c081c80915f8378 2020-03-31T20:00:00Z \n", + "1 06fc31f03af09c0f09f4c5bc18fac9daa63e94f8 2020-03-31T20:00:00Z \n", "\n", " totalTestResults fips deathIncrease hospitalizedIncrease \\\n", - "0 3654 2 1.0 1.0 \n", - "1 6553 1 2.0 0.0 \n", + "0 3713 2 0.0 0.0 \n", + "1 7279 1 7.0 0.0 \n", "\n", " negativeIncrease positiveIncrease totalTestResultsIncrease \n", - "0 308.0 12.0 320.0 \n", - "1 1510.0 53.0 1563.0 " + "0 54.0 5.0 59.0 \n", + "1 604.0 122.0 726.0 " ] }, "execution_count": 9, @@ -455,18 +455,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 3.290092, - "end_time": "2020-03-31T08:24:32.748032", + "duration": 2.364995, + "end_time": "2020-04-01T11:24:58.291712", "environment_variables": {}, "exception": null, - "input_path": "/tmp/anascddl/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/v_x6sgie/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/anascddl/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/v_x6sgie/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-03-31T08:24:29.457940", + "start_time": "2020-04-01T11:24:55.926717", "version": "1.1.0" } },