diff --git a/.renku/workflow/efbdf4faf6d44153b72a3640f38d4831.cwl b/.renku/workflow/efbdf4faf6d44153b72a3640f38d4831.cwl new file mode 100644 index 0000000000000000000000000000000000000000..299afccddd5c837b398d8df8303e7a19c99d8e34 --- /dev/null +++ b/.renku/workflow/efbdf4faf6d44153b72a3640f38d4831.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: out_folder + streamable: false + type: string + input_2: + default: data/covidtracking + streamable: false + type: string + input_3: + default: + class: File + path: ../../notebooks/process/download-covidtracking-data.ipynb + streamable: false + type: File + input_4: + default: runs/download-covidtracking-data.runs.ipynb + streamable: false + type: string + input_5: + default: states-daily.json + streamable: false + type: string + input_6: + default: states-metadata.json + streamable: false + type: string +outputs: + output_1: + outputSource: step_1/output_1 + streamable: false + type: Directory + output_3: + outputSource: step_1/output_0 + streamable: false + type: File +requirements: [] +steps: + step_1: + in: + input_1: input_1 + input_2: input_2 + input_3: input_3 + input_4: input_4 + out: + - output_0 + - output_1 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl + step_2: + in: + filename: input_5 + input_directory: step_1/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 + class: InlineJavascriptRequirement + - &id002 + class: InitialWorkDirRequirement + listing: $(inputs.input_directory.listing) + successCodes: [] + temporaryFailCodes: [] + step_3: + in: + filename: input_6 + input_directory: step_1/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 + - *id002 + successCodes: [] + temporaryFailCodes: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 956bc691b809b6452e4b084745fb5b69aea660e3..313b4dcd8ba026393bf8d856e6f3ed39fb0fbff2 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 sha256:8b56cd3e63adc5312bee8ff1200f683bf0eedb6d3f2a8062e14bf14427bfb380 -size 20235966 +oid sha256:3e3f573dfadd19add61f23f173c691376a9e7314f675c8c95d4d6587250003de +size 20320781 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 6755c3dd306f8b980931902cdb3867f243059f34..338e4649aefc02038004ea48c352bed36bd9c059 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 sha256:1e5d2f5ddd46de871293f810366044abbcd262d74f31f0e6c902fa4296762d9b -size 128145 +oid sha256:ae36642b16f7dc2df25eba3634adfd5772731de2874d7bd625fe1b87d8a8bed7 +size 129621 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index a8eb5e15ee17d03325f75d01c6f4656c13d12f28..5e9efabf370f5c19464dfb67b3ed22b72c8639fb 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": 4.315171, - "end_time": "2020-11-23T07:38:05.453200", + "duration": 5.255276, + "end_time": "2020-11-24T07:44:49.459961", "exception": false, - "start_time": "2020-11-23T07:38:01.138029", + "start_time": "2020-11-24T07:44:44.204685", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.021568, - "end_time": "2020-11-23T07:38:05.490145", + "duration": 0.023889, + "end_time": "2020-11-24T07:44:49.499512", "exception": false, - "start_time": "2020-11-23T07:38:05.468577", + "start_time": "2020-11-24T07:44:49.475623", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.016979, - "end_time": "2020-11-23T07:38:05.515643", + "duration": 0.010921, + "end_time": "2020-11-24T07:44:49.516597", "exception": false, - "start_time": "2020-11-23T07:38:05.498664", + "start_time": "2020-11-24T07:44:49.505676", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/gti3qup1/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/t0y_essg/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": 0.006709, - "end_time": "2020-11-23T07:38:05.527788", + "duration": 0.004343, + "end_time": "2020-11-24T07:44:49.525362", "exception": false, - "start_time": "2020-11-23T07:38:05.521079", + "start_time": "2020-11-24T07:44:49.521019", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.819824, - "end_time": "2020-11-23T07:38:06.352491", + "duration": 10.391367, + "end_time": "2020-11-24T07:44:59.920943", "exception": false, - "start_time": "2020-11-23T07:38:05.532667", + "start_time": "2020-11-24T07:44:49.529576", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.024026, - "end_time": "2020-11-23T07:38:06.388073", + "duration": 0.024074, + "end_time": "2020-11-24T07:44:59.957871", "exception": false, - "start_time": "2020-11-23T07:38:06.364047", + "start_time": "2020-11-24T07:44:59.933797", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.071082, - "end_time": "2020-11-23T07:38:06.468790", + "duration": 0.066306, + "end_time": "2020-11-24T07:45:00.032836", "exception": false, - "start_time": "2020-11-23T07:38:06.397708", + "start_time": "2020-11-24T07:44:59.966530", "status": "completed" }, "tags": [] @@ -284,10 +284,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.00659, - "end_time": "2020-11-23T07:38:06.486563", + "duration": 0.007437, + "end_time": "2020-11-24T07:45:00.052559", "exception": false, - "start_time": "2020-11-23T07:38:06.479973", + "start_time": "2020-11-24T07:45:00.045122", "status": "completed" }, "tags": [] @@ -301,10 +301,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.631132, - "end_time": "2020-11-23T07:38:07.123740", + "duration": 10.634454, + "end_time": "2020-11-24T07:45:10.693263", "exception": false, - "start_time": "2020-11-23T07:38:06.492608", + "start_time": "2020-11-24T07:45:00.058809", "status": "completed" }, "tags": [] @@ -321,10 +321,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.048362, - "end_time": "2020-11-23T07:38:07.186789", + "duration": 0.047788, + "end_time": "2020-11-24T07:45:10.754629", "exception": false, - "start_time": "2020-11-23T07:38:07.138427", + "start_time": "2020-11-24T07:45:10.706841", "status": "completed" }, "tags": [] @@ -343,10 +343,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.5637, - "end_time": "2020-11-23T07:38:07.764924", + "duration": 0.55723, + "end_time": "2020-11-24T07:45:11.324807", "exception": false, - "start_time": "2020-11-23T07:38:07.201224", + "start_time": "2020-11-24T07:45:10.767577", "status": "completed" }, "tags": [] @@ -356,7 +356,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "14841 data points\n" + "14899 data points\n" ] }, { @@ -406,21 +406,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201122</td>\n", + " <td>20201123</td>\n", " <td>AK</td>\n", - " <td>26587.0</td>\n", + " <td>27085.0</td>\n", " <td>NaN</td>\n", - " <td>899774.0</td>\n", + " <td>905931.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>926361.0</td>\n", - " <td>133.0</td>\n", - " <td>612.0</td>\n", + " <td>933016.0</td>\n", + " <td>146.0</td>\n", + " <td>617.0</td>\n", " <td>...</td>\n", - " <td>926361</td>\n", - " <td>1</td>\n", + " <td>933016</td>\n", " <td>0</td>\n", - " <td>597df917af4cf6d52bb240230523f16f37ab083e</td>\n", + " <td>5</td>\n", + " <td>cbbe56430899e87f4ea8a0336bdc01fdb3b34998</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>20201122</td>\n", + " <td>20201123</td>\n", " <td>AL</td>\n", - " <td>232506.0</td>\n", - " <td>37921.0</td>\n", - " <td>1330801.0</td>\n", + " <td>234080.0</td>\n", + " <td>38193.0</td>\n", + " <td>1337741.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1525386.0</td>\n", - " <td>1332.0</td>\n", - " <td>23449.0</td>\n", + " <td>1533628.0</td>\n", + " <td>1427.0</td>\n", + " <td>23942.0</td>\n", " <td>...</td>\n", - " <td>1563307</td>\n", - " <td>0</td>\n", - " <td>0</td>\n", - " <td>547eed4fa2ca857a0b99b1bc8989877ce9200b22</td>\n", + " <td>1571821</td>\n", + " <td>2</td>\n", + " <td>493</td>\n", + " <td>064b957eb226a47780acf699e8e184c49ced247b</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 20201122 AK 26587.0 NaN 899774.0 NaN \n", - "1 20201122 AL 232506.0 37921.0 1330801.0 NaN \n", + "0 20201123 AK 27085.0 NaN 905931.0 NaN \n", + "1 20201123 AL 234080.0 38193.0 1337741.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 926361.0 133.0 \n", - "1 totalTestsViral 1525386.0 1332.0 \n", + "0 totalTestsViral 933016.0 146.0 \n", + "1 totalTestsViral 1533628.0 1427.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 612.0 ... 926361 1 0 \n", - "1 23449.0 ... 1563307 0 0 \n", + "0 617.0 ... 933016 0 5 \n", + "1 23942.0 ... 1571821 2 493 \n", "\n", " hash commercialScore \\\n", - "0 597df917af4cf6d52bb240230523f16f37ab083e 0 \n", - "1 547eed4fa2ca857a0b99b1bc8989877ce9200b22 0 \n", + "0 cbbe56430899e87f4ea8a0336bdc01fdb3b34998 0 \n", + "1 064b957eb226a47780acf699e8e184c49ced247b 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": 9.070785, - "end_time": "2020-11-23T07:38:08.088519", + "duration": 29.443362, + "end_time": "2020-11-24T07:45:11.646275", "environment_variables": {}, "exception": null, - "input_path": "/tmp/gti3qup1/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/t0y_essg/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/gti3qup1/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/t0y_essg/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-23T07:37:59.017734", + "start_time": "2020-11-24T07:44:42.202913", "version": "1.1.0" } },