diff --git a/.renku/workflow/20c86d3d7885417b805f91afd932a627.cwl b/.renku/workflow/20c86d3d7885417b805f91afd932a627.cwl new file mode 100644 index 0000000000000000000000000000000000000000..a8e35b68b1209ea481386bf236383679302cee0b --- /dev/null +++ b/.renku/workflow/20c86d3d7885417b805f91afd932a627.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_1 + streamable: false + type: Directory + output_1: + 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 1be11444642e4f974ac224f70975e82fec993ad0..ce3fa2695f12f9b0422fac3d113a07d005108b7c 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 sha256:8c1ab862b0edc7d2ec86cb1515e0ef920fa2ce0c5fd5b4a7fd928e9c54c8868e -size 8291546 +oid sha256:5acd8695ad7773e3784bbabfb54e3e6a3379c968894f5c37e545b88d92bd676b +size 8344111 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 1394cf3fd636295c7c3611c72b850bed9ac7cb61..9c523018f35660e2eb6afa8ae6fc8726a55ce041 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 sha256:5eb6e8452ea20eac3ffa209e980ad6bc8feedcc70773855978038d614336ef2a -size 59497 +oid sha256:4ce25de8d46e9a0c7f96e7ab57cbc6fcf3a943ff2fd37a183678b83f428f36ec +size 60597 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 3a6a7edd3d2602bf2bd8350b7a6a429d14635676..b9510f70402cb5c1738a99a8f3775328e6623b47 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": 4.909303, - "end_time": "2020-08-12T07:32:52.940667", + "duration": 2.138843, + "end_time": "2020-08-13T07:32:08.076759", "exception": false, - "start_time": "2020-08-12T07:32:48.031364", + "start_time": "2020-08-13T07:32:05.937916", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.020786, - "end_time": "2020-08-12T07:32:52.974199", + "duration": 0.019514, + "end_time": "2020-08-13T07:32:08.109425", "exception": false, - "start_time": "2020-08-12T07:32:52.953413", + "start_time": "2020-08-13T07:32:08.089911", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.018808, - "end_time": "2020-08-12T07:32:53.002848", + "duration": 0.020546, + "end_time": "2020-08-13T07:32:08.138868", "exception": false, - "start_time": "2020-08-12T07:32:52.984040", + "start_time": "2020-08-13T07:32:08.118322", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/fa34vlpr/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/8ktvuu43/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": 0.007114, - "end_time": "2020-08-12T07:32:53.017421", + "duration": 0.007322, + "end_time": "2020-08-13T07:32:08.155663", "exception": false, - "start_time": "2020-08-12T07:32:53.010307", + "start_time": "2020-08-13T07:32:08.148341", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.253919, - "end_time": "2020-08-12T07:32:53.278771", + "duration": 0.288666, + "end_time": "2020-08-13T07:32:08.451980", "exception": false, - "start_time": "2020-08-12T07:32:53.024852", + "start_time": "2020-08-13T07:32:08.163314", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.023702, - "end_time": "2020-08-12T07:32:53.315338", + "duration": 0.021771, + "end_time": "2020-08-13T07:32:08.485568", "exception": false, - "start_time": "2020-08-12T07:32:53.291636", + "start_time": "2020-08-13T07:32:08.463797", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.112023, - "end_time": "2020-08-12T07:32:53.437535", + "duration": 0.075526, + "end_time": "2020-08-13T07:32:08.570959", "exception": false, - "start_time": "2020-08-12T07:32:53.325512", + "start_time": "2020-08-13T07:32:08.495433", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.010012, - "end_time": "2020-08-12T07:32:53.458945", + "duration": 0.008029, + "end_time": "2020-08-13T07:32:08.591772", "exception": false, - "start_time": "2020-08-12T07:32:53.448933", + "start_time": "2020-08-13T07:32:08.583743", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.847674, - "end_time": "2020-08-12T07:32:54.316583", + "duration": 0.912442, + "end_time": "2020-08-13T07:32:09.512304", "exception": false, - "start_time": "2020-08-12T07:32:53.468909", + "start_time": "2020-08-13T07:32:08.599862", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.034853, - "end_time": "2020-08-12T07:32:54.363904", + "duration": 0.032626, + "end_time": "2020-08-13T07:32:09.560190", "exception": false, - "start_time": "2020-08-12T07:32:54.329051", + "start_time": "2020-08-13T07:32:09.527564", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.359212, - "end_time": "2020-08-12T07:32:54.737206", + "duration": 0.363787, + "end_time": "2020-08-13T07:32:09.935148", "exception": false, - "start_time": "2020-08-12T07:32:54.377994", + "start_time": "2020-08-13T07:32:09.571361", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "8921 data points\n" + "8977 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200811</td>\n", + " <td>20200812</td>\n", " <td>AK</td>\n", - " <td>4589.0</td>\n", - " <td>286155.0</td>\n", + " <td>4655.0</td>\n", + " <td>287927.0</td>\n", " <td>NaN</td>\n", " <td>39.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", - " <td>4.0</td>\n", + " <td>3.0</td>\n", " <td>...</td>\n", - " <td>290744</td>\n", - " <td>0</td>\n", + " <td>292582</td>\n", + " <td>1</td>\n", " <td>0</td>\n", - " <td>75189f7e45df3d19bc61fc8790a6df68e33c41e4</td>\n", + " <td>04d3f681722d466f15660e1ded53888693484520</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -395,21 +395,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200811</td>\n", + " <td>20200812</td>\n", " <td>AL</td>\n", - " <td>103851.0</td>\n", - " <td>679838.0</td>\n", + " <td>104786.0</td>\n", + " <td>690985.0</td>\n", " <td>NaN</td>\n", - " <td>1506.0</td>\n", - " <td>12070.0</td>\n", + " <td>1372.0</td>\n", + " <td>12292.0</td>\n", " <td>NaN</td>\n", - " <td>1270.0</td>\n", + " <td>1282.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>783689</td>\n", - " <td>50</td>\n", - " <td>0</td>\n", - " <td>d775daf03cf949c25340db800bb25f4eec3c5fda</td>\n", + " <td>795771</td>\n", + " <td>35</td>\n", + " <td>222</td>\n", + " <td>3b8e75d37adb9ba3951ec8b2aae85ba1fa9b02f1</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -424,20 +424,20 @@ ], "text/plain": [ " date state positive negative pending hospitalizedCurrently \\\n", - "0 20200811 AK 4589.0 286155.0 NaN 39.0 \n", - "1 20200811 AL 103851.0 679838.0 NaN 1506.0 \n", + "0 20200812 AK 4655.0 287927.0 NaN 39.0 \n", + "1 20200812 AL 104786.0 690985.0 NaN 1372.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 12070.0 NaN 1270.0 \n", + "1 12292.0 NaN 1282.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 4.0 ... 290744 0 0 \n", - "1 NaN ... 783689 50 0 \n", + "0 3.0 ... 292582 1 0 \n", + "1 NaN ... 795771 35 222 \n", "\n", " hash commercialScore \\\n", - "0 75189f7e45df3d19bc61fc8790a6df68e33c41e4 0 \n", - "1 d775daf03cf949c25340db800bb25f4eec3c5fda 0 \n", + "0 04d3f681722d466f15660e1ded53888693484520 0 \n", + "1 3b8e75d37adb9ba3951ec8b2aae85ba1fa9b02f1 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -477,18 +477,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 8.638524, - "end_time": "2020-08-12T07:32:55.057745", + "duration": 5.465167, + "end_time": "2020-08-13T07:32:10.255042", "environment_variables": {}, "exception": null, - "input_path": "/tmp/fa34vlpr/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/8ktvuu43/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/fa34vlpr/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/8ktvuu43/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-08-12T07:32:46.419221", + "start_time": "2020-08-13T07:32:04.789875", "version": "1.1.0" } },