diff --git a/.renku/workflow/95900e3b5544477fa36891fe80b71e6a.cwl b/.renku/workflow/95900e3b5544477fa36891fe80b71e6a.cwl new file mode 100644 index 0000000000000000000000000000000000000000..b611ac623c5d5e6d0474e0bbd43bb2fa2581deef --- /dev/null +++ b/.renku/workflow/95900e3b5544477fa36891fe80b71e6a.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_1: + 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 451bde44e36499342a97eae629aec58f0f98a7b1..2f095ef2fd735027e9f3c9f814347edd3edc0852 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 sha256:0e93551218f4f81392d24f6ff17007b5121303ee9eb8dbc1559da7bec33c93b2 -size 7659954 +oid sha256:4a21497c9679604fb3a884b4621c552a77c070b108edf4d11427545366900bf5 +size 7712627 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index e561fa9622cfa022eda83539c390478cb223efc6..66087951a2b07b53031cc9d4ffe76fc1d9dfe823 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 sha256:97f32e3b08ceb5019605543e0a90d8b53c0933ad0950677c58fd868943bd2f38 -size 56338 +oid sha256:7bdcbea46addae2bd6392349d71b66545282603a1f4362689f95ef337a232937 +size 57221 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 9d0290c1e293d065698885606d17b3adc6af9559..66a52f842934a53ddcdeb638c118fb416e2d5dac 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": 0.636339, - "end_time": "2020-07-31T07:31:38.916840", + "duration": 3.677738, + "end_time": "2020-08-01T07:32:36.765169", "exception": false, - "start_time": "2020-07-31T07:31:38.280501", + "start_time": "2020-08-01T07:32:33.087431", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.017811, - "end_time": "2020-07-31T07:31:38.947335", + "duration": 0.024832, + "end_time": "2020-08-01T07:32:36.806906", "exception": false, - "start_time": "2020-07-31T07:31:38.929524", + "start_time": "2020-08-01T07:32:36.782074", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.019772, - "end_time": "2020-07-31T07:31:38.974996", + "duration": 0.022172, + "end_time": "2020-08-01T07:32:36.839469", "exception": false, - "start_time": "2020-07-31T07:31:38.955224", + "start_time": "2020-08-01T07:32:36.817297", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/z_m8yogb/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/xpfpy1ny/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": 0.006274, - "end_time": "2020-07-31T07:31:38.988972", + "duration": 0.00615, + "end_time": "2020-08-01T07:32:36.854666", "exception": false, - "start_time": "2020-07-31T07:31:38.982698", + "start_time": "2020-08-01T07:32:36.848516", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.31111, - "end_time": "2020-07-31T07:31:39.306628", + "duration": 0.184244, + "end_time": "2020-08-01T07:32:37.044786", "exception": false, - "start_time": "2020-07-31T07:31:38.995518", + "start_time": "2020-08-01T07:32:36.860542", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.019968, - "end_time": "2020-07-31T07:31:39.336696", + "duration": 0.032584, + "end_time": "2020-08-01T07:32:37.098311", "exception": false, - "start_time": "2020-07-31T07:31:39.316728", + "start_time": "2020-08-01T07:32:37.065727", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.058807, - "end_time": "2020-07-31T07:31:39.403115", + "duration": 0.076098, + "end_time": "2020-08-01T07:32:37.188818", "exception": false, - "start_time": "2020-07-31T07:31:39.344308", + "start_time": "2020-08-01T07:32:37.112720", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007059, - "end_time": "2020-07-31T07:31:39.417766", + "duration": 0.00836, + "end_time": "2020-08-01T07:32:37.211485", "exception": false, - "start_time": "2020-07-31T07:31:39.410707", + "start_time": "2020-08-01T07:32:37.203125", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.234612, - "end_time": "2020-07-31T07:31:39.659826", + "duration": 0.24849, + "end_time": "2020-08-01T07:32:37.467781", "exception": false, - "start_time": "2020-07-31T07:31:39.425214", + "start_time": "2020-08-01T07:32:37.219291", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.029778, - "end_time": "2020-07-31T07:31:39.701277", + "duration": 0.033347, + "end_time": "2020-08-01T07:32:37.518652", "exception": false, - "start_time": "2020-07-31T07:31:39.671499", + "start_time": "2020-08-01T07:32:37.485305", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.346106, - "end_time": "2020-07-31T07:31:40.058722", + "duration": 0.307833, + "end_time": "2020-08-01T07:32:37.837752", "exception": false, - "start_time": "2020-07-31T07:31:39.712616", + "start_time": "2020-08-01T07:32:37.529919", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "8249 data points\n" + "8305 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200730</td>\n", + " <td>20200731</td>\n", " <td>AK</td>\n", - " <td>3559.0</td>\n", - " <td>221498.0</td>\n", + " <td>3677.0</td>\n", + " <td>229429.0</td>\n", " <td>NaN</td>\n", - " <td>45.0</td>\n", + " <td>40.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>3.0</td>\n", " <td>...</td>\n", - " <td>225057</td>\n", - " <td>1</td>\n", + " <td>233106</td>\n", + " <td>0</td>\n", " <td>0</td>\n", - " <td>d66445af73028a02a5d6713580a4c470cc603322</td>\n", + " <td>435bc6d4686226ba3a508751129be8b2ef723836</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>20200730</td>\n", + " <td>20200731</td>\n", " <td>AL</td>\n", - " <td>85762.0</td>\n", - " <td>594346.0</td>\n", + " <td>87723.0</td>\n", + " <td>601744.0</td>\n", " <td>NaN</td>\n", - " <td>1595.0</td>\n", - " <td>10070.0</td>\n", + " <td>1596.0</td>\n", + " <td>10521.0</td>\n", " <td>NaN</td>\n", - " <td>1129.0</td>\n", + " <td>1144.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>680108</td>\n", - " <td>27</td>\n", - " <td>177</td>\n", - " <td>5776f9724a3452e48705a9b2c73f52d3e631229d</td>\n", + " <td>689467</td>\n", + " <td>15</td>\n", + " <td>451</td>\n", + " <td>310a029041a6b87a179110674657ee90aac288c0</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 20200730 AK 3559.0 221498.0 NaN 45.0 \n", - "1 20200730 AL 85762.0 594346.0 NaN 1595.0 \n", + "0 20200731 AK 3677.0 229429.0 NaN 40.0 \n", + "1 20200731 AL 87723.0 601744.0 NaN 1596.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 10070.0 NaN 1129.0 \n", + "1 10521.0 NaN 1144.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 3.0 ... 225057 1 0 \n", - "1 NaN ... 680108 27 177 \n", + "0 3.0 ... 233106 0 0 \n", + "1 NaN ... 689467 15 451 \n", "\n", " hash commercialScore \\\n", - "0 d66445af73028a02a5d6713580a4c470cc603322 0 \n", - "1 5776f9724a3452e48705a9b2c73f52d3e631229d 0 \n", + "0 435bc6d4686226ba3a508751129be8b2ef723836 0 \n", + "1 310a029041a6b87a179110674657ee90aac288c0 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": 3.000062, - "end_time": "2020-07-31T07:31:40.379070", + "duration": 6.732796, + "end_time": "2020-08-01T07:32:38.164781", "environment_variables": {}, "exception": null, - "input_path": "/tmp/z_m8yogb/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/xpfpy1ny/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/z_m8yogb/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/xpfpy1ny/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-07-31T07:31:37.379008", + "start_time": "2020-08-01T07:32:31.431985", "version": "1.1.0" } },