diff --git a/.renku/workflow/4b986549374544639a21ec2269579141.cwl b/.renku/workflow/4b986549374544639a21ec2269579141.cwl new file mode 100644 index 0000000000000000000000000000000000000000..08c8c8255c5f108bccae2c8c7fe108a7c4ab64f1 --- /dev/null +++ b/.renku/workflow/4b986549374544639a21ec2269579141.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: states-daily.json + streamable: false + type: string + input_2: + default: states-metadata.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_2: + 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-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_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-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: [] + 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 305d9455a5e79f4eccfeab6a9c4946c71008a9af..323c9bc1fee720ad02d570e1184c5ea6ce62a2fc 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 sha256:43110b45fbcaf372e87ef2da0578a7738d41ab1ef43c88163f61b52cd26d0c8f -size 18921545 +oid sha256:c7ecdf2d64be4aa613b3159d8d7135e3794ed9d6cf5996465c2e9aa67de471ae +size 18998763 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index cb98443aaed7f2b829f385f7e5c8d25e64c9e317..6637cf0544cef0071cd3d4196fc34c0ed661ffb7 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 sha256:3856ce65b8cdf051d3887ddb14a1fed8c0ca94fa3028146ea446348f9ad7eec0 -size 112431 +oid sha256:f5ec84c44d9e948f34339a61f69a8f5fab844e8d0cb6824767001c8abeb475f9 +size 113826 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index b0388faba7dbf1839ae0a8d498f2bad54faa9a25..2d849b57508c0c0506bcddd2901b34f642fccd71 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": 4.567062, - "end_time": "2020-11-06T07:36:16.535139", + "duration": 4.17924, + "end_time": "2020-11-07T07:36:26.642620", "exception": false, - "start_time": "2020-11-06T07:36:11.968077", + "start_time": "2020-11-07T07:36:22.463380", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.021675, - "end_time": "2020-11-06T07:36:16.571363", + "duration": 0.020794, + "end_time": "2020-11-07T07:36:26.676440", "exception": false, - "start_time": "2020-11-06T07:36:16.549688", + "start_time": "2020-11-07T07:36:26.655646", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.016527, - "end_time": "2020-11-06T07:36:16.595797", + "duration": 0.018287, + "end_time": "2020-11-07T07:36:26.702242", "exception": false, - "start_time": "2020-11-06T07:36:16.579270", + "start_time": "2020-11-07T07:36:26.683955", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/cvg3ldzb/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/q9ded2bw/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": 0.004867, - "end_time": "2020-11-06T07:36:16.605495", + "duration": 0.005456, + "end_time": "2020-11-07T07:36:26.713871", "exception": false, - "start_time": "2020-11-06T07:36:16.600628", + "start_time": "2020-11-07T07:36:26.708415", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.679735, - "end_time": "2020-11-06T07:36:17.289998", + "duration": 0.677261, + "end_time": "2020-11-07T07:36:27.395390", "exception": false, - "start_time": "2020-11-06T07:36:16.610263", + "start_time": "2020-11-07T07:36:26.718129", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.022148, - "end_time": "2020-11-06T07:36:17.323825", + "duration": 0.02652, + "end_time": "2020-11-07T07:36:27.435303", "exception": false, - "start_time": "2020-11-06T07:36:17.301677", + "start_time": "2020-11-07T07:36:27.408783", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.124039, - "end_time": "2020-11-06T07:36:17.456351", + "duration": 0.071178, + "end_time": "2020-11-07T07:36:27.518446", "exception": false, - "start_time": "2020-11-06T07:36:17.332312", + "start_time": "2020-11-07T07:36:27.447268", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.006929, - "end_time": "2020-11-06T07:36:17.474962", + "duration": 0.00671, + "end_time": "2020-11-07T07:36:27.536537", "exception": false, - "start_time": "2020-11-06T07:36:17.468033", + "start_time": "2020-11-07T07:36:27.529827", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.608193, - "end_time": "2020-11-06T07:36:18.089165", + "duration": 0.457032, + "end_time": "2020-11-07T07:36:27.999709", "exception": false, - "start_time": "2020-11-06T07:36:17.480972", + "start_time": "2020-11-07T07:36:27.542677", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.041932, - "end_time": "2020-11-06T07:36:18.146875", + "duration": 0.045366, + "end_time": "2020-11-07T07:36:28.059656", "exception": false, - "start_time": "2020-11-06T07:36:18.104943", + "start_time": "2020-11-07T07:36:28.014290", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.55035, - "end_time": "2020-11-06T07:36:18.711417", + "duration": 0.545213, + "end_time": "2020-11-07T07:36:28.617366", "exception": false, - "start_time": "2020-11-06T07:36:18.161067", + "start_time": "2020-11-07T07:36:28.072153", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13887 data points\n" + "13943 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201105</td>\n", + " <td>20201106</td>\n", " <td>AK</td>\n", - " <td>18174.0</td>\n", + " <td>18699.0</td>\n", " <td>NaN</td>\n", - " <td>763652.0</td>\n", + " <td>762602.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>781826.0</td>\n", - " <td>104.0</td>\n", + " <td>781301.0</td>\n", + " <td>106.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>781826</td>\n", + " <td>781301</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>910e0e68ec1831715444823aa43d30e203d4ec90</td>\n", + " <td>c5e7641b5775740f5912af833d4d6b5c452c3383</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>20201105</td>\n", + " <td>20201106</td>\n", " <td>AL</td>\n", - " <td>199158.0</td>\n", - " <td>29892.0</td>\n", - " <td>1209841.0</td>\n", + " <td>200714.0</td>\n", + " <td>30268.0</td>\n", + " <td>1217002.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1379107.0</td>\n", - " <td>994.0</td>\n", - " <td>21027.0</td>\n", + " <td>1387448.0</td>\n", + " <td>1022.0</td>\n", + " <td>21294.0</td>\n", " <td>...</td>\n", - " <td>1408999</td>\n", - " <td>20</td>\n", - " <td>0</td>\n", - " <td>e660184b515c15867ae651a03b8c7f2f23271e52</td>\n", + " <td>1417716</td>\n", + " <td>23</td>\n", + " <td>267</td>\n", + " <td>3c28ba3b012f38332635f3629ab4fa30bddaf663</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 20201105 AK 18174.0 NaN 763652.0 NaN \n", - "1 20201105 AL 199158.0 29892.0 1209841.0 NaN \n", + "0 20201106 AK 18699.0 NaN 762602.0 NaN \n", + "1 20201106 AL 200714.0 30268.0 1217002.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 781826.0 104.0 \n", - "1 totalTestsViral 1379107.0 994.0 \n", + "0 totalTestsViral 781301.0 106.0 \n", + "1 totalTestsViral 1387448.0 1022.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 781826 0 0 \n", - "1 21027.0 ... 1408999 20 0 \n", + "0 NaN ... 781301 0 0 \n", + "1 21294.0 ... 1417716 23 267 \n", "\n", " hash commercialScore \\\n", - "0 910e0e68ec1831715444823aa43d30e203d4ec90 0 \n", - "1 e660184b515c15867ae651a03b8c7f2f23271e52 0 \n", + "0 c5e7641b5775740f5912af833d4d6b5c452c3383 0 \n", + "1 3c28ba3b012f38332635f3629ab4fa30bddaf663 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": 8.490223, - "end_time": "2020-11-06T07:36:19.033810", + "duration": 8.158202, + "end_time": "2020-11-07T07:36:28.941551", "environment_variables": {}, "exception": null, - "input_path": "/tmp/cvg3ldzb/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/q9ded2bw/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/cvg3ldzb/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/q9ded2bw/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-06T07:36:10.543587", + "start_time": "2020-11-07T07:36:20.783349", "version": "1.1.0" } },