diff --git a/.renku/workflow/050e8f458e484c0281c4846de954afc4.cwl b/.renku/workflow/050e8f458e484c0281c4846de954afc4.cwl new file mode 100644 index 0000000000000000000000000000000000000000..a0067321b319ba052a8214144f686dadf2c8ec12 --- /dev/null +++ b/.renku/workflow/050e8f458e484c0281c4846de954afc4.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_1: + outputSource: step_2/output_1 + streamable: false + type: Directory + output_2: + outputSource: step_2/output_0 + streamable: false + type: File +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 89e38926d251824a8368c05303803c7cbfb33a96..305d9455a5e79f4eccfeab6a9c4946c71008a9af 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 sha256:2ed324b60f9234bd8aac0bf8bb578d96458dddd114bd13441a39ca9a958d660b -size 18844434 +oid sha256:43110b45fbcaf372e87ef2da0578a7738d41ab1ef43c88163f61b52cd26d0c8f +size 18921545 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index e804187c31ca72ae796a0b5e6b1ff34a4a826577..cb98443aaed7f2b829f385f7e5c8d25e64c9e317 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 sha256:64019b2ea484dda369afd8063c139dbd29e791776e41a66cda0c4e36adaa2ae4 -size 110301 +oid sha256:3856ce65b8cdf051d3887ddb14a1fed8c0ca94fa3028146ea446348f9ad7eec0 +size 112431 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index bdad21c41e32e363a655a00b6b8b0e4e3d3de556..b0388faba7dbf1839ae0a8d498f2bad54faa9a25 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": 6.729895, - "end_time": "2020-11-05T07:38:41.062106", + "duration": 4.567062, + "end_time": "2020-11-06T07:36:16.535139", "exception": false, - "start_time": "2020-11-05T07:38:34.332211", + "start_time": "2020-11-06T07:36:11.968077", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.024761, - "end_time": "2020-11-05T07:38:41.103587", + "duration": 0.021675, + "end_time": "2020-11-06T07:36:16.571363", "exception": false, - "start_time": "2020-11-05T07:38:41.078826", + "start_time": "2020-11-06T07:36:16.549688", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.026788, - "end_time": "2020-11-05T07:38:41.142628", + "duration": 0.016527, + "end_time": "2020-11-06T07:36:16.595797", "exception": false, - "start_time": "2020-11-05T07:38:41.115840", + "start_time": "2020-11-06T07:36:16.579270", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/vbj6b6ni/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/cvg3ldzb/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": 0.009469, - "end_time": "2020-11-05T07:38:41.167225", + "duration": 0.004867, + "end_time": "2020-11-06T07:36:16.605495", "exception": false, - "start_time": "2020-11-05T07:38:41.157756", + "start_time": "2020-11-06T07:36:16.600628", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.242981, - "end_time": "2020-11-05T07:38:41.417207", + "duration": 0.679735, + "end_time": "2020-11-06T07:36:17.289998", "exception": false, - "start_time": "2020-11-05T07:38:41.174226", + "start_time": "2020-11-06T07:36:16.610263", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.030029, - "end_time": "2020-11-05T07:38:41.462901", + "duration": 0.022148, + "end_time": "2020-11-06T07:36:17.323825", "exception": false, - "start_time": "2020-11-05T07:38:41.432872", + "start_time": "2020-11-06T07:36:17.301677", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.111829, - "end_time": "2020-11-05T07:38:41.587925", + "duration": 0.124039, + "end_time": "2020-11-06T07:36:17.456351", "exception": false, - "start_time": "2020-11-05T07:38:41.476096", + "start_time": "2020-11-06T07:36:17.332312", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007981, - "end_time": "2020-11-05T07:38:41.608573", + "duration": 0.006929, + "end_time": "2020-11-06T07:36:17.474962", "exception": false, - "start_time": "2020-11-05T07:38:41.600592", + "start_time": "2020-11-06T07:36:17.468033", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.452865, - "end_time": "2020-11-05T07:38:42.068098", + "duration": 0.608193, + "end_time": "2020-11-06T07:36:18.089165", "exception": false, - "start_time": "2020-11-05T07:38:41.615233", + "start_time": "2020-11-06T07:36:17.480972", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.049196, - "end_time": "2020-11-05T07:38:42.134677", + "duration": 0.041932, + "end_time": "2020-11-06T07:36:18.146875", "exception": false, - "start_time": "2020-11-05T07:38:42.085481", + "start_time": "2020-11-06T07:36:18.104943", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.643402, - "end_time": "2020-11-05T07:38:42.793243", + "duration": 0.55035, + "end_time": "2020-11-06T07:36:18.711417", "exception": false, - "start_time": "2020-11-05T07:38:42.149841", + "start_time": "2020-11-06T07:36:18.161067", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13831 data points\n" + "13887 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201104</td>\n", + " <td>20201105</td>\n", " <td>AK</td>\n", - " <td>17860.0</td>\n", + " <td>18174.0</td>\n", " <td>NaN</td>\n", - " <td>609898.0</td>\n", + " <td>763652.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>627758.0</td>\n", - " <td>95.0</td>\n", + " <td>781826.0</td>\n", + " <td>104.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>627758</td>\n", + " <td>781826</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>b1f3faafdb8100fd0d67eb31a3614c20da9f076c</td>\n", + " <td>910e0e68ec1831715444823aa43d30e203d4ec90</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>20201104</td>\n", + " <td>20201105</td>\n", " <td>AL</td>\n", - " <td>197777.0</td>\n", - " <td>29662.0</td>\n", - " <td>1203497.0</td>\n", + " <td>199158.0</td>\n", + " <td>29892.0</td>\n", + " <td>1209841.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1371612.0</td>\n", - " <td>1017.0</td>\n", + " <td>1379107.0</td>\n", + " <td>994.0</td>\n", " <td>21027.0</td>\n", " <td>...</td>\n", - " <td>1401274</td>\n", - " <td>19</td>\n", - " <td>126</td>\n", - " <td>1bac7f8b3ff7ab5f7136284a077b2ab1fd62cc5b</td>\n", + " <td>1408999</td>\n", + " <td>20</td>\n", + " <td>0</td>\n", + " <td>e660184b515c15867ae651a03b8c7f2f23271e52</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 20201104 AK 17860.0 NaN 609898.0 NaN \n", - "1 20201104 AL 197777.0 29662.0 1203497.0 NaN \n", + "0 20201105 AK 18174.0 NaN 763652.0 NaN \n", + "1 20201105 AL 199158.0 29892.0 1209841.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 627758.0 95.0 \n", - "1 totalTestsViral 1371612.0 1017.0 \n", + "0 totalTestsViral 781826.0 104.0 \n", + "1 totalTestsViral 1379107.0 994.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 627758 0 0 \n", - "1 21027.0 ... 1401274 19 126 \n", + "0 NaN ... 781826 0 0 \n", + "1 21027.0 ... 1408999 20 0 \n", "\n", " hash commercialScore \\\n", - "0 b1f3faafdb8100fd0d67eb31a3614c20da9f076c 0 \n", - "1 1bac7f8b3ff7ab5f7136284a077b2ab1fd62cc5b 0 \n", + "0 910e0e68ec1831715444823aa43d30e203d4ec90 0 \n", + "1 e660184b515c15867ae651a03b8c7f2f23271e52 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": 10.85433, - "end_time": "2020-11-05T07:38:43.223805", + "duration": 8.490223, + "end_time": "2020-11-06T07:36:19.033810", "environment_variables": {}, "exception": null, - "input_path": "/tmp/vbj6b6ni/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/cvg3ldzb/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/vbj6b6ni/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/cvg3ldzb/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-05T07:38:32.369475", + "start_time": "2020-11-06T07:36:10.543587", "version": "1.1.0" } },