diff --git a/.renku/workflow/812230b2beb242f38b032a29e4483a9f.cwl b/.renku/workflow/812230b2beb242f38b032a29e4483a9f.cwl new file mode 100644 index 0000000000000000000000000000000000000000..0e74f62c250e0a5761b975f82a0aaaedd400c645 --- /dev/null +++ b/.renku/workflow/812230b2beb242f38b032a29e4483a9f.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: states-metadata.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-daily.json + streamable: false + type: string +outputs: + output_0: + 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-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 + 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_0 + - output_1 + 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-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 + - *id002 + successCodes: [] + temporaryFailCodes: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 1555ddfcf0f135f467e7516bf9c1e034106b30e3..d6c801a243689ce3a37a6f0b74874698759115e5 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:f7073509fa6668976fd62235a4e4b6bf0da964fb1656fff9738e99f4ea6c69b3 -size 7554680 +oid sha256:97c134a06483128db08e938cf71875392cbea85979fbd26c74781fce36af065d +size 7607316 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 2160d9b58e778907bb3486ab5be16a829564785a..3884a502a5a30449a5699a77f91a39272a98d21f 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:c00124de21c434a218eed723fb751d1b2b6c4dd294118e95836ff189ebdbc9cd -size 55384 +oid sha256:05e5ccb1f970e3a175428e24ab590fb9617b6758099708f6e6059a48e6140aa9 +size 56087 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 4a91fa1c326bf01715f36932352ee9ff8440f1dd..e5b552504c59d81c8896277c6e7a9619e9dd6a9e 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": 1.634067, - "end_time": "2020-07-29T07:32:51.603280", + "duration": 4.245417, + "end_time": "2020-07-30T07:31:17.545891", "exception": false, - "start_time": "2020-07-29T07:32:49.969213", + "start_time": "2020-07-30T07:31:13.300474", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.027599, - "end_time": "2020-07-29T07:32:51.645061", + "duration": 0.017034, + "end_time": "2020-07-30T07:31:17.576851", "exception": false, - "start_time": "2020-07-29T07:32:51.617462", + "start_time": "2020-07-30T07:31:17.559817", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.022819, - "end_time": "2020-07-29T07:32:51.676504", + "duration": 0.020583, + "end_time": "2020-07-30T07:31:17.604547", "exception": false, - "start_time": "2020-07-29T07:32:51.653685", + "start_time": "2020-07-30T07:31:17.583964", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/0mrm7jst/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/vg71jqgt/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": 0.009235, - "end_time": "2020-07-29T07:32:51.700436", + "duration": 0.007165, + "end_time": "2020-07-30T07:31:17.621588", "exception": false, - "start_time": "2020-07-29T07:32:51.691201", + "start_time": "2020-07-30T07:31:17.614423", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.082633, - "end_time": "2020-07-29T07:32:51.789179", + "duration": 0.172965, + "end_time": "2020-07-30T07:31:17.801988", "exception": false, - "start_time": "2020-07-29T07:32:51.706546", + "start_time": "2020-07-30T07:31:17.629023", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.026774, - "end_time": "2020-07-29T07:32:51.838420", + "duration": 0.019319, + "end_time": "2020-07-30T07:31:17.833009", "exception": false, - "start_time": "2020-07-29T07:32:51.811646", + "start_time": "2020-07-30T07:31:17.813690", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.09229, - "end_time": "2020-07-29T07:32:51.941010", + "duration": 0.066186, + "end_time": "2020-07-30T07:31:17.908024", "exception": false, - "start_time": "2020-07-29T07:32:51.848720", + "start_time": "2020-07-30T07:31:17.841838", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007278, - "end_time": "2020-07-29T07:32:51.960099", + "duration": 0.008214, + "end_time": "2020-07-30T07:31:17.927031", "exception": false, - "start_time": "2020-07-29T07:32:51.952821", + "start_time": "2020-07-30T07:31:17.918817", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.313721, - "end_time": "2020-07-29T07:32:52.281007", + "duration": 0.326219, + "end_time": "2020-07-30T07:31:18.261696", "exception": false, - "start_time": "2020-07-29T07:32:51.967286", + "start_time": "2020-07-30T07:31:17.935477", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.035029, - "end_time": "2020-07-29T07:32:52.331598", + "duration": 0.025568, + "end_time": "2020-07-30T07:31:18.299253", "exception": false, - "start_time": "2020-07-29T07:32:52.296569", + "start_time": "2020-07-30T07:31:18.273685", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.368739, - "end_time": "2020-07-29T07:32:52.713521", + "duration": 0.331916, + "end_time": "2020-07-30T07:31:18.639298", "exception": false, - "start_time": "2020-07-29T07:32:52.344782", + "start_time": "2020-07-30T07:31:18.307382", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "8137 data points\n" + "8193 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200728</td>\n", + " <td>20200729</td>\n", " <td>AK</td>\n", - " <td>3350.0</td>\n", - " <td>210525.0</td>\n", + " <td>3442.0</td>\n", + " <td>220222.0</td>\n", " <td>NaN</td>\n", - " <td>44.0</td>\n", + " <td>45.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", - " <td>4.0</td>\n", + " <td>3.0</td>\n", " <td>...</td>\n", - " <td>213875</td>\n", - " <td>1</td>\n", + " <td>223664</td>\n", + " <td>0</td>\n", " <td>0</td>\n", - " <td>ae5f917caa2078779c9ac0c581af0b3740885be7</td>\n", + " <td>e3c05df1cf92b9c80941cec8daaf723905e7a0dd</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -395,21 +395,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200728</td>\n", + " <td>20200729</td>\n", " <td>AL</td>\n", - " <td>82366.0</td>\n", - " <td>583033.0</td>\n", + " <td>83782.0</td>\n", + " <td>584493.0</td>\n", " <td>NaN</td>\n", " <td>1598.0</td>\n", " <td>9893.0</td>\n", " <td>NaN</td>\n", - " <td>1094.0</td>\n", + " <td>1109.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>665399</td>\n", + " <td>668275</td>\n", + " <td>47</td>\n", " <td>0</td>\n", - " <td>199</td>\n", - " <td>c1a5bd8f18d98fc9530cd541a50772d21cbfb38a</td>\n", + " <td>a7b049cfd14780e56e6ef98fb2aba9f82fe89838</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -424,20 +424,20 @@ ], "text/plain": [ " date state positive negative pending hospitalizedCurrently \\\n", - "0 20200728 AK 3350.0 210525.0 NaN 44.0 \n", - "1 20200728 AL 82366.0 583033.0 NaN 1598.0 \n", + "0 20200729 AK 3442.0 220222.0 NaN 45.0 \n", + "1 20200729 AL 83782.0 584493.0 NaN 1598.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 9893.0 NaN 1094.0 \n", + "1 9893.0 NaN 1109.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 4.0 ... 213875 1 0 \n", - "1 NaN ... 665399 0 199 \n", + "0 3.0 ... 223664 0 0 \n", + "1 NaN ... 668275 47 0 \n", "\n", " hash commercialScore \\\n", - "0 ae5f917caa2078779c9ac0c581af0b3740885be7 0 \n", - "1 c1a5bd8f18d98fc9530cd541a50772d21cbfb38a 0 \n", + "0 e3c05df1cf92b9c80941cec8daaf723905e7a0dd 0 \n", + "1 a7b049cfd14780e56e6ef98fb2aba9f82fe89838 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -477,18 +477,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 4.508495, - "end_time": "2020-07-29T07:32:53.140366", + "duration": 7.336817, + "end_time": "2020-07-30T07:31:18.963931", "environment_variables": {}, "exception": null, - "input_path": "/tmp/0mrm7jst/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/vg71jqgt/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/0mrm7jst/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/vg71jqgt/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-07-29T07:32:48.631871", + "start_time": "2020-07-30T07:31:11.627114", "version": "1.1.0" } },