diff --git a/.renku/workflow/dbdecdb499f04eee86d54e69d32cba44.cwl b/.renku/workflow/dbdecdb499f04eee86d54e69d32cba44.cwl new file mode 100644 index 0000000000000000000000000000000000000000..bf3d45b32b4d5c704c364f99318a2acf1fdc743e --- /dev/null +++ b/.renku/workflow/dbdecdb499f04eee86d54e69d32cba44.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_1 + streamable: false + type: Directory + output_2: + outputSource: step_3/output_0 + streamable: false + type: File +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_1 + - output_0 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index cdfc89c699dc3163797eef25bed7389d6857db86..4599fa881e3862839816aed69d5d622295aebd92 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 sha256:9efebd725653a363e81e3ce23049783f07a69fc11c7baf53b4180832e074c705 -size 16904054 +oid sha256:f3e6ca4a0e573fb7c61646da3830e19767e4d6e572dabf41d2e37d455f7e62d2 +size 16980722 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 4275d0d934b67741c1962b4c44329caaa0194b56..86b2da912a1b8e618b9c1c732bf2b3892a7a360f 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 sha256:221fe0ffc46c51469e093fee5b79437b97dcefde8de4dac5b4e2ec4415aa8953 -size 95250 +oid sha256:819cc5dfeac04b86836b9fab05a39b13a545e06e9cd3fb5b9b300438608bad29 +size 95977 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index b248fdc78bbc61ce0043041399d27df23ded3314..5afe8f82bc12ed407dedc3734ba30d99f63de87d 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": 4.367513, - "end_time": "2020-10-13T07:37:36.058414", + "duration": 2.358472, + "end_time": "2020-10-14T07:36:57.770956", "exception": false, - "start_time": "2020-10-13T07:37:31.690901", + "start_time": "2020-10-14T07:36:55.412484", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.024929, - "end_time": "2020-10-13T07:37:36.100157", + "duration": 0.024658, + "end_time": "2020-10-14T07:36:57.811834", "exception": false, - "start_time": "2020-10-13T07:37:36.075228", + "start_time": "2020-10-14T07:36:57.787176", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.022783, - "end_time": "2020-10-13T07:37:36.135278", + "duration": 0.021832, + "end_time": "2020-10-14T07:36:57.845563", "exception": false, - "start_time": "2020-10-13T07:37:36.112495", + "start_time": "2020-10-14T07:36:57.823731", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/0u0vlt9d/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/wkiajyw0/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": 0.005511, - "end_time": "2020-10-13T07:37:36.148935", + "duration": 0.005956, + "end_time": "2020-10-14T07:36:57.859365", "exception": false, - "start_time": "2020-10-13T07:37:36.143424", + "start_time": "2020-10-14T07:36:57.853409", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.783371, - "end_time": "2020-10-13T07:37:36.939210", + "duration": 0.961671, + "end_time": "2020-10-14T07:36:58.826923", "exception": false, - "start_time": "2020-10-13T07:37:36.155839", + "start_time": "2020-10-14T07:36:57.865252", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.027963, - "end_time": "2020-10-13T07:37:36.980895", + "duration": 0.024675, + "end_time": "2020-10-14T07:36:58.867816", "exception": false, - "start_time": "2020-10-13T07:37:36.952932", + "start_time": "2020-10-14T07:36:58.843141", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.08441, - "end_time": "2020-10-13T07:37:37.077147", + "duration": 0.116794, + "end_time": "2020-10-14T07:36:58.992950", "exception": false, - "start_time": "2020-10-13T07:37:36.992737", + "start_time": "2020-10-14T07:36:58.876156", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008393, - "end_time": "2020-10-13T07:37:37.099679", + "duration": 0.008136, + "end_time": "2020-10-14T07:36:59.013531", "exception": false, - "start_time": "2020-10-13T07:37:37.091286", + "start_time": "2020-10-14T07:36:59.005395", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 27.922089, - "end_time": "2020-10-13T07:38:05.028680", + "duration": 28.308817, + "end_time": "2020-10-14T07:37:27.328847", "exception": false, - "start_time": "2020-10-13T07:37:37.106591", + "start_time": "2020-10-14T07:36:59.020030", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.054229, - "end_time": "2020-10-13T07:38:05.101847", + "duration": 0.032088, + "end_time": "2020-10-14T07:37:27.369420", "exception": false, - "start_time": "2020-10-13T07:38:05.047618", + "start_time": "2020-10-14T07:37:27.337332", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.602416, - "end_time": "2020-10-13T07:38:05.720664", + "duration": 0.568302, + "end_time": "2020-10-14T07:37:27.944447", "exception": false, - "start_time": "2020-10-13T07:38:05.118248", + "start_time": "2020-10-14T07:37:27.376145", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "12477 data points\n" + "12533 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201012</td>\n", + " <td>20201013</td>\n", " <td>AK</td>\n", - " <td>10884.0</td>\n", + " <td>11039.0</td>\n", " <td>NaN</td>\n", - " <td>495043.0</td>\n", + " <td>503750.0</td>\n", " <td>NaN</td>\n", - " <td>505927.0</td>\n", - " <td>50.0</td>\n", + " <td>514789.0</td>\n", + " <td>59.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>505927</td>\n", + " <td>514789</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>1d12a6644f8d97132e890998a17090d44e70d4e1</td>\n", + " <td>881fbdb23caeb53c47e5007c8599ec1624e1c68d</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>20201012</td>\n", + " <td>20201013</td>\n", " <td>AL</td>\n", - " <td>166076.0</td>\n", - " <td>19853.0</td>\n", - " <td>1068387.0</td>\n", + " <td>167193.0</td>\n", + " <td>20110.0</td>\n", + " <td>1074124.0</td>\n", " <td>NaN</td>\n", - " <td>1214610.0</td>\n", - " <td>856.0</td>\n", - " <td>18179.0</td>\n", + " <td>1221207.0</td>\n", + " <td>823.0</td>\n", + " <td>18440.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>1234463</td>\n", - " <td>1</td>\n", + " <td>1241317</td>\n", " <td>0</td>\n", - " <td>feb6ff2653368253706dc08cae65080d91c5d680</td>\n", + " <td>261</td>\n", + " <td>19ea68f7f6d374d276ce094b7473ceb24e9e0cae</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 20201012 AK 10884.0 NaN 495043.0 NaN \n", - "1 20201012 AL 166076.0 19853.0 1068387.0 NaN \n", + "0 20201013 AK 11039.0 NaN 503750.0 NaN \n", + "1 20201013 AL 167193.0 20110.0 1074124.0 NaN \n", "\n", " totalTestResults hospitalizedCurrently hospitalizedCumulative \\\n", - "0 505927.0 50.0 NaN \n", - "1 1214610.0 856.0 18179.0 \n", + "0 514789.0 59.0 NaN \n", + "1 1221207.0 823.0 18440.0 \n", "\n", " inIcuCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 505927 0 0 \n", - "1 NaN ... 1234463 1 0 \n", + "0 NaN ... 514789 0 0 \n", + "1 NaN ... 1241317 0 261 \n", "\n", " hash commercialScore \\\n", - "0 1d12a6644f8d97132e890998a17090d44e70d4e1 0 \n", - "1 feb6ff2653368253706dc08cae65080d91c5d680 0 \n", + "0 881fbdb23caeb53c47e5007c8599ec1624e1c68d 0 \n", + "1 19ea68f7f6d374d276ce094b7473ceb24e9e0cae 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": 36.408658, - "end_time": "2020-10-13T07:38:06.155114", + "duration": 34.640425, + "end_time": "2020-10-14T07:37:28.363129", "environment_variables": {}, "exception": null, - "input_path": "/tmp/0u0vlt9d/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/wkiajyw0/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/0u0vlt9d/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/wkiajyw0/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-10-13T07:37:29.746456", + "start_time": "2020-10-14T07:36:53.722704", "version": "1.1.0" } },