diff --git a/.renku/workflow/00f12aa777134609bc85411e3f5d8f6b.cwl b/.renku/workflow/00f12aa777134609bc85411e3f5d8f6b.cwl new file mode 100644 index 0000000000000000000000000000000000000000..3e1c0278af0680e92eb03adcd381585981dd97c6 --- /dev/null +++ b/.renku/workflow/00f12aa777134609bc85411e3f5d8f6b.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: states-daily.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_0 + streamable: false + type: File + output_3: + 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-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: + 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-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: [] + 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 7f38a010c11347bf97e64bd4c44fcd40155471bb..8d53d0c2ab8d4154bf3ce4af71f21484697edd9e 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 sha256:16682d903ed8a6b0ccf4ca43afd355faa7b7ce5aba1c5b503222b90c5303496e -size 17818101 +oid sha256:51ad395a5bf2b0dde015a25045dd5d982591167b21bdbf99d72a51085935e0fe +size 17895030 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 0362e76fce0d3e4eccba7d2ce2ce593d4caa02b5..2f4c82cc7503a4dba103f43a4460560059e5e553 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:f90646fc532386199074033947a8e0e0cff4125da706f1a1dc99e9ddcf3f8095 -size 102172 +oid sha256:462decf550f00953608ce2ea38092635fbe8b563a67081eb916f96311be18acc +size 102176 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index d568f10637bfba5ee2130831536ced411fbe2230..5784883f0bb6f2ef54d5baa57b610eb4c5ad7274 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": 3.520049, - "end_time": "2020-10-24T07:53:17.756990", + "duration": 3.550558, + "end_time": "2020-10-25T07:37:45.875174", "exception": false, - "start_time": "2020-10-24T07:53:14.236941", + "start_time": "2020-10-25T07:37:42.324616", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.027464, - "end_time": "2020-10-24T07:53:17.801073", + "duration": 0.027824, + "end_time": "2020-10-25T07:37:45.918328", "exception": false, - "start_time": "2020-10-24T07:53:17.773609", + "start_time": "2020-10-25T07:37:45.890504", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.02419, - "end_time": "2020-10-24T07:53:17.836826", + "duration": 0.027753, + "end_time": "2020-10-25T07:37:45.960330", "exception": false, - "start_time": "2020-10-24T07:53:17.812636", + "start_time": "2020-10-25T07:37:45.932577", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/ni4orh0_/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/g156d8nk/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": 0.007658, - "end_time": "2020-10-24T07:53:17.852212", + "duration": 0.007933, + "end_time": "2020-10-25T07:37:45.980254", "exception": false, - "start_time": "2020-10-24T07:53:17.844554", + "start_time": "2020-10-25T07:37:45.972321", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.617007, - "end_time": "2020-10-24T07:53:18.474980", + "duration": 0.6145, + "end_time": "2020-10-25T07:37:46.602858", "exception": false, - "start_time": "2020-10-24T07:53:17.857973", + "start_time": "2020-10-25T07:37:45.988358", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.029508, - "end_time": "2020-10-24T07:53:18.520364", + "duration": 0.019979, + "end_time": "2020-10-25T07:37:46.634561", "exception": false, - "start_time": "2020-10-24T07:53:18.490856", + "start_time": "2020-10-25T07:37:46.614582", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.075423, - "end_time": "2020-10-24T07:53:18.606156", + "duration": 0.088566, + "end_time": "2020-10-25T07:37:46.735021", "exception": false, - "start_time": "2020-10-24T07:53:18.530733", + "start_time": "2020-10-25T07:37:46.646455", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008246, - "end_time": "2020-10-24T07:53:18.628357", + "duration": 0.007662, + "end_time": "2020-10-25T07:37:46.755490", "exception": false, - "start_time": "2020-10-24T07:53:18.620111", + "start_time": "2020-10-25T07:37:46.747828", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 29.719818, - "end_time": "2020-10-24T07:53:48.355673", + "duration": 29.73858, + "end_time": "2020-10-25T07:38:16.502066", "exception": false, - "start_time": "2020-10-24T07:53:18.635855", + "start_time": "2020-10-25T07:37:46.763486", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.067044, - "end_time": "2020-10-24T07:53:48.440245", + "duration": 0.064526, + "end_time": "2020-10-25T07:38:16.581680", "exception": false, - "start_time": "2020-10-24T07:53:48.373201", + "start_time": "2020-10-25T07:38:16.517154", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.658594, - "end_time": "2020-10-24T07:53:49.114333", + "duration": 0.658515, + "end_time": "2020-10-25T07:38:17.252086", "exception": false, - "start_time": "2020-10-24T07:53:48.455739", + "start_time": "2020-10-25T07:38:16.593571", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13101 data points\n" + "13157 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201023</td>\n", + " <td>20201024</td>\n", " <td>AK</td>\n", - " <td>13161.0</td>\n", + " <td>13535.0</td>\n", " <td>NaN</td>\n", " <td>539585.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", " <td>552746.0</td>\n", - " <td>59.0</td>\n", + " <td>58.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>552746</td>\n", + " <td>553120</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>1b73fb16c097754ce2024af0e49f9d9514248af0</td>\n", + " <td>280ee400bd797c20b77218c9e54a0b6615f91a53</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>20201023</td>\n", + " <td>20201024</td>\n", " <td>AL</td>\n", - " <td>180916.0</td>\n", - " <td>25001.0</td>\n", - " <td>1133858.0</td>\n", + " <td>183276.0</td>\n", + " <td>26330.0</td>\n", + " <td>1138922.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1289773.0</td>\n", - " <td>888.0</td>\n", + " <td>1295868.0</td>\n", + " <td>920.0</td>\n", " <td>19595.0</td>\n", " <td>...</td>\n", - " <td>1314774</td>\n", - " <td>16</td>\n", - " <td>147</td>\n", - " <td>5323772280706f0dd467c842b6640d6f931d3c5a</td>\n", + " <td>1322198</td>\n", + " <td>7</td>\n", + " <td>0</td>\n", + " <td>7c16e74ce103ea23b8b2ef6881fc773d44d21f55</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 20201023 AK 13161.0 NaN 539585.0 NaN \n", - "1 20201023 AL 180916.0 25001.0 1133858.0 NaN \n", + "0 20201024 AK 13535.0 NaN 539585.0 NaN \n", + "1 20201024 AL 183276.0 26330.0 1138922.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 552746.0 59.0 \n", - "1 totalTestsViral 1289773.0 888.0 \n", + "0 totalTestsViral 552746.0 58.0 \n", + "1 totalTestsViral 1295868.0 920.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 552746 0 0 \n", - "1 19595.0 ... 1314774 16 147 \n", + "0 NaN ... 553120 0 0 \n", + "1 19595.0 ... 1322198 7 0 \n", "\n", " hash commercialScore \\\n", - "0 1b73fb16c097754ce2024af0e49f9d9514248af0 0 \n", - "1 5323772280706f0dd467c842b6640d6f931d3c5a 0 \n", + "0 280ee400bd797c20b77218c9e54a0b6615f91a53 0 \n", + "1 7c16e74ce103ea23b8b2ef6881fc773d44d21f55 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": 37.048735, - "end_time": "2020-10-24T07:53:49.543034", + "duration": 36.766249, + "end_time": "2020-10-25T07:38:17.574642", "environment_variables": {}, "exception": null, - "input_path": "/tmp/ni4orh0_/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/g156d8nk/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/ni4orh0_/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/g156d8nk/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-10-24T07:53:12.494299", + "start_time": "2020-10-25T07:37:40.808393", "version": "1.1.0" } },