diff --git a/.renku/workflow/7d6191ace0bd4c75bfa6debfc195b857.cwl b/.renku/workflow/7d6191ace0bd4c75bfa6debfc195b857.cwl new file mode 100644 index 0000000000000000000000000000000000000000..4954d051a85321c882c22370c097db11e7654adc --- /dev/null +++ b/.renku/workflow/7d6191ace0bd4c75bfa6debfc195b857.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_0: + outputSource: step_1/output_0 + streamable: false + type: File + output_2: + outputSource: step_1/output_1 + streamable: false + type: Directory +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 3b31fd5e4dfc8f0a58bdec607bab3e20f01b7739..a6c08f854787af8fbe4082a9a01c2f1a51e6a8d0 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 sha256:40c4e7c36a3a3edadce9aa9741dca93fda9bb696d41ffe83157f69afc1a45604 -size 18459097 +oid sha256:62d83a9e7e219b75a90a189b1afb7bc9b6bf64b0c1640f0a71d25ac2395c0017 +size 18536189 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index dd134035858a3a69df7aa93fad234bc9bf26480c..e377d225be631104357677458ab4407b6e03fb34 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:bb5e9f61edebb173aea0a7c02035ee98d39365bd0b3fb95c513697e502342668 -size 108377 +oid sha256:e5c9982f4d81279d78cee5958a09ef5197dde0a9ac1d96cc2cb456bf4629da9d +size 108379 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 9a3388ebb4868c0c9b3fa1525cc670a1e9d37220..014eaa71d48963607ccf2f7efd7ccbdb2f261827 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": 3.682234, - "end_time": "2020-10-31T07:41:00.543129", + "duration": 5.037062, + "end_time": "2020-11-01T07:37:45.933708", "exception": false, - "start_time": "2020-10-31T07:40:56.860895", + "start_time": "2020-11-01T07:37:40.896646", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.024558, - "end_time": "2020-10-31T07:41:00.583002", + "duration": 0.02828, + "end_time": "2020-11-01T07:37:45.978315", "exception": false, - "start_time": "2020-10-31T07:41:00.558444", + "start_time": "2020-11-01T07:37:45.950035", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.019837, - "end_time": "2020-10-31T07:41:00.610773", + "duration": 0.022512, + "end_time": "2020-11-01T07:37:46.013706", "exception": false, - "start_time": "2020-10-31T07:41:00.590936", + "start_time": "2020-11-01T07:37:45.991194", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/5go_iu__/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/y_sxckx7/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": 0.006231, - "end_time": "2020-10-31T07:41:00.624248", + "duration": 0.006401, + "end_time": "2020-11-01T07:37:46.029106", "exception": false, - "start_time": "2020-10-31T07:41:00.618017", + "start_time": "2020-11-01T07:37:46.022705", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.859229, - "end_time": "2020-10-31T07:41:01.488799", + "duration": 0.65798, + "end_time": "2020-11-01T07:37:46.693110", "exception": false, - "start_time": "2020-10-31T07:41:00.629570", + "start_time": "2020-11-01T07:37:46.035130", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.030144, - "end_time": "2020-10-31T07:41:01.534355", + "duration": 0.028713, + "end_time": "2020-11-01T07:37:46.735011", "exception": false, - "start_time": "2020-10-31T07:41:01.504211", + "start_time": "2020-11-01T07:37:46.706298", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.099661, - "end_time": "2020-10-31T07:41:01.644716", + "duration": 0.090248, + "end_time": "2020-11-01T07:37:46.838574", "exception": false, - "start_time": "2020-10-31T07:41:01.545055", + "start_time": "2020-11-01T07:37:46.748326", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007944, - "end_time": "2020-10-31T07:41:01.667560", + "duration": 0.007958, + "end_time": "2020-11-01T07:37:46.859707", "exception": false, - "start_time": "2020-10-31T07:41:01.659616", + "start_time": "2020-11-01T07:37:46.851749", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 30.643978, - "end_time": "2020-10-31T07:41:32.318512", + "duration": 0.437219, + "end_time": "2020-11-01T07:37:47.303780", "exception": false, - "start_time": "2020-10-31T07:41:01.674534", + "start_time": "2020-11-01T07:37:46.866561", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.054579, - "end_time": "2020-10-31T07:41:32.390496", + "duration": 0.063124, + "end_time": "2020-11-01T07:37:47.378855", "exception": false, - "start_time": "2020-10-31T07:41:32.335917", + "start_time": "2020-11-01T07:37:47.315731", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.594878, - "end_time": "2020-10-31T07:41:33.000128", + "duration": 0.650152, + "end_time": "2020-11-01T07:37:48.043379", "exception": false, - "start_time": "2020-10-31T07:41:32.405250", + "start_time": "2020-11-01T07:37:47.393227", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13551 data points\n" + "13607 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201030</td>\n", + " <td>20201031</td>\n", " <td>AK</td>\n", - " <td>15908.0</td>\n", + " <td>16355.0</td>\n", " <td>NaN</td>\n", - " <td>570415.0</td>\n", + " <td>587852.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>586323.0</td>\n", - " <td>90.0</td>\n", + " <td>604207.0</td>\n", + " <td>94.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>586323</td>\n", - " <td>4</td>\n", + " <td>604207</td>\n", + " <td>1</td>\n", " <td>0</td>\n", - " <td>f02fcc28ab30329793653f2d5759c257699cb0de</td>\n", + " <td>4fd1e9352fd403f75cd0f4b8c7d420de8eb73f0b</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>20201030</td>\n", + " <td>20201031</td>\n", " <td>AL</td>\n", - " <td>190496.0</td>\n", - " <td>27776.0</td>\n", - " <td>1172969.0</td>\n", + " <td>192285.0</td>\n", + " <td>27990.0</td>\n", + " <td>1181958.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1335689.0</td>\n", - " <td>1013.0</td>\n", + " <td>1346253.0</td>\n", + " <td>960.0</td>\n", " <td>20450.0</td>\n", " <td>...</td>\n", - " <td>1363465</td>\n", - " <td>18</td>\n", - " <td>204</td>\n", - " <td>2731923ac8ca3480f0dfe4eb973a4345f06e5643</td>\n", + " <td>1374243</td>\n", + " <td>35</td>\n", + " <td>0</td>\n", + " <td>5194e1ea04ad4dace9ab68daca52f9fa98e39f1b</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 20201030 AK 15908.0 NaN 570415.0 NaN \n", - "1 20201030 AL 190496.0 27776.0 1172969.0 NaN \n", + "0 20201031 AK 16355.0 NaN 587852.0 NaN \n", + "1 20201031 AL 192285.0 27990.0 1181958.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 586323.0 90.0 \n", - "1 totalTestsViral 1335689.0 1013.0 \n", + "0 totalTestsViral 604207.0 94.0 \n", + "1 totalTestsViral 1346253.0 960.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 586323 4 0 \n", - "1 20450.0 ... 1363465 18 204 \n", + "0 NaN ... 604207 1 0 \n", + "1 20450.0 ... 1374243 35 0 \n", "\n", " hash commercialScore \\\n", - "0 f02fcc28ab30329793653f2d5759c257699cb0de 0 \n", - "1 2731923ac8ca3480f0dfe4eb973a4345f06e5643 0 \n", + "0 4fd1e9352fd403f75cd0f4b8c7d420de8eb73f0b 0 \n", + "1 5194e1ea04ad4dace9ab68daca52f9fa98e39f1b 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": 38.867596, - "end_time": "2020-10-31T07:41:33.426697", + "duration": 9.337475, + "end_time": "2020-11-01T07:37:48.472569", "environment_variables": {}, "exception": null, - "input_path": "/tmp/5go_iu__/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/y_sxckx7/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/5go_iu__/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/y_sxckx7/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-10-31T07:40:54.559101", + "start_time": "2020-11-01T07:37:39.135094", "version": "1.1.0" } },