diff --git a/.renku/workflow/dc2813e1ceab42f0b02443dfa079dd05.cwl b/.renku/workflow/dc2813e1ceab42f0b02443dfa079dd05.cwl new file mode 100644 index 0000000000000000000000000000000000000000..1add8a8c398a9ad2e9e59a4f3ee00777aee562d0 --- /dev/null +++ b/.renku/workflow/dc2813e1ceab42f0b02443dfa079dd05.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_1: + outputSource: step_2/output_0 + streamable: false + type: File + output_3: + outputSource: step_2/output_1 + streamable: false + type: Directory +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_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-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 3e31bc293a44899a2fa3f2aa59d148d34bc81f78..bdb759f47ed9caf804bde6c6ead757710af0f86b 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 sha256:6834cb04ee15ed650960ef5b8d6f14b8fb6b801f46fd0d8b7970b1d163750d72 -size 21802399 +oid sha256:f6cf5cdba96fd179d0327beb6dec5f4dc2acdf7be4e4254d60a870ee58ff7973 +size 21879715 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 2d34465ee50ac74a42a8fb81770dfc9243ad64e7..ee4e7cd99a83525d9b125e8b4bb98493de56b6de 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 sha256:7354c67941de362dc05fa040ab2f436300cb1d766c501671e16643a447f1ec46 -size 151830 +oid sha256:7cc9e0b090ce7ee03d0dc6e8ffc87ce22e82979dc9522306f179f15c89647eee +size 151994 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 7a1003c031cc136780a6f4dc51cd239c23130e3d..04207bc8bad9bc15a0d509091609a6d036fa9554 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": 5.223516, - "end_time": "2020-12-13T07:37:44.521551", + "duration": 6.566808, + "end_time": "2020-12-14T07:38:54.524277", "exception": false, - "start_time": "2020-12-13T07:37:39.298035", + "start_time": "2020-12-14T07:38:47.957469", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.020038, - "end_time": "2020-12-13T07:37:44.555668", + "duration": 0.021334, + "end_time": "2020-12-14T07:38:54.559337", "exception": false, - "start_time": "2020-12-13T07:37:44.535630", + "start_time": "2020-12-14T07:38:54.538003", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.016584, - "end_time": "2020-12-13T07:37:44.579810", + "duration": 0.018262, + "end_time": "2020-12-14T07:38:54.585141", "exception": false, - "start_time": "2020-12-13T07:37:44.563226", + "start_time": "2020-12-14T07:38:54.566879", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/_2i_io6h/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/x_0_3ys0/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": 0.005697, - "end_time": "2020-12-13T07:37:44.591532", + "duration": 0.005928, + "end_time": "2020-12-14T07:38:54.597606", "exception": false, - "start_time": "2020-12-13T07:37:44.585835", + "start_time": "2020-12-14T07:38:54.591678", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 1.048541, - "end_time": "2020-12-13T07:37:45.645366", + "duration": 0.880372, + "end_time": "2020-12-14T07:38:55.482495", "exception": false, - "start_time": "2020-12-13T07:37:44.596825", + "start_time": "2020-12-14T07:38:54.602123", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.024147, - "end_time": "2020-12-13T07:37:45.682788", + "duration": 0.024586, + "end_time": "2020-12-14T07:38:55.524191", "exception": false, - "start_time": "2020-12-13T07:37:45.658641", + "start_time": "2020-12-14T07:38:55.499605", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.083844, - "end_time": "2020-12-13T07:37:45.779457", + "duration": 0.072201, + "end_time": "2020-12-14T07:38:55.602915", "exception": false, - "start_time": "2020-12-13T07:37:45.695613", + "start_time": "2020-12-14T07:38:55.530714", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008011, - "end_time": "2020-12-13T07:37:45.801291", + "duration": 0.007188, + "end_time": "2020-12-14T07:38:55.621955", "exception": false, - "start_time": "2020-12-13T07:37:45.793280", + "start_time": "2020-12-14T07:38:55.614767", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.679409, - "end_time": "2020-12-13T07:37:46.487312", + "duration": 1.000764, + "end_time": "2020-12-14T07:38:56.628208", "exception": false, - "start_time": "2020-12-13T07:37:45.807903", + "start_time": "2020-12-14T07:38:55.627444", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.046951, - "end_time": "2020-12-13T07:37:46.549317", + "duration": 0.059235, + "end_time": "2020-12-14T07:38:56.703467", "exception": false, - "start_time": "2020-12-13T07:37:46.502366", + "start_time": "2020-12-14T07:38:56.644232", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.607161, - "end_time": "2020-12-13T07:37:47.168953", + "duration": 0.584686, + "end_time": "2020-12-14T07:38:57.301657", "exception": false, - "start_time": "2020-12-13T07:37:46.561792", + "start_time": "2020-12-14T07:38:56.716971", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "15969 data points\n" + "16025 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201212</td>\n", + " <td>20201213</td>\n", " <td>AK</td>\n", - " <td>39101.0</td>\n", + " <td>39738.0</td>\n", " <td>NaN</td>\n", - " <td>1094317.0</td>\n", + " <td>1103360.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1133418.0</td>\n", - " <td>146.0</td>\n", - " <td>883.0</td>\n", + " <td>1143098.0</td>\n", + " <td>139.0</td>\n", + " <td>886.0</td>\n", " <td>...</td>\n", - " <td>1133418</td>\n", - " <td>18</td>\n", - " <td>14</td>\n", - " <td>a2d18ec064bf83b8ac5c9f93a2e0e38827d0865b</td>\n", + " <td>1143098</td>\n", + " <td>0</td>\n", + " <td>3</td>\n", + " <td>d156168dc751b7d5ac6c800bb113036257e330ac</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -430,21 +430,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20201212</td>\n", + " <td>20201213</td>\n", " <td>AL</td>\n", - " <td>292841.0</td>\n", - " <td>52306.0</td>\n", + " <td>295631.0</td>\n", + " <td>52801.0</td>\n", " <td>1458560.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsPeopleViral</td>\n", " <td>1696439.0</td>\n", - " <td>2161.0</td>\n", + " <td>2248.0</td>\n", " <td>28146.0</td>\n", " <td>...</td>\n", - " <td>1751401</td>\n", - " <td>16</td>\n", + " <td>1754191</td>\n", + " <td>0</td>\n", " <td>0</td>\n", - " <td>660f90fe723b05c91856029bfd46d43423c107e2</td>\n", + " <td>4e514bc9880b07c124076d06e42235040016b194</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -459,20 +459,20 @@ ], "text/plain": [ " date state positive probableCases negative pending \\\n", - "0 20201212 AK 39101.0 NaN 1094317.0 NaN \n", - "1 20201212 AL 292841.0 52306.0 1458560.0 NaN \n", + "0 20201213 AK 39738.0 NaN 1103360.0 NaN \n", + "1 20201213 AL 295631.0 52801.0 1458560.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 1133418.0 146.0 \n", - "1 totalTestsPeopleViral 1696439.0 2161.0 \n", + "0 totalTestsViral 1143098.0 139.0 \n", + "1 totalTestsPeopleViral 1696439.0 2248.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 883.0 ... 1133418 18 14 \n", - "1 28146.0 ... 1751401 16 0 \n", + "0 886.0 ... 1143098 0 3 \n", + "1 28146.0 ... 1754191 0 0 \n", "\n", " hash commercialScore \\\n", - "0 a2d18ec064bf83b8ac5c9f93a2e0e38827d0865b 0 \n", - "1 660f90fe723b05c91856029bfd46d43423c107e2 0 \n", + "0 d156168dc751b7d5ac6c800bb113036257e330ac 0 \n", + "1 4e514bc9880b07c124076d06e42235040016b194 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -512,18 +512,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 9.926778, - "end_time": "2020-12-13T07:37:47.490053", + "duration": 13.185672, + "end_time": "2020-12-14T07:38:57.625906", "environment_variables": {}, "exception": null, - "input_path": "/tmp/_2i_io6h/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/x_0_3ys0/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/_2i_io6h/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/x_0_3ys0/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-12-13T07:37:37.563275", + "start_time": "2020-12-14T07:38:44.440234", "version": "1.1.0" } },