diff --git a/.renku/workflow/0a1ea867681b42e9babe7bd954cfcf54.cwl b/.renku/workflow/0a1ea867681b42e9babe7bd954cfcf54.cwl new file mode 100644 index 0000000000000000000000000000000000000000..de4a095f84fb3800ed9630517dbdb7ee9f0e874e --- /dev/null +++ b/.renku/workflow/0a1ea867681b42e9babe7bd954cfcf54.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_2: + 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 b7b4847ba1fc445c41ad9cefbd5ab923ee49ece2..48da09c3cbff59e3e01fea220c41b8a7ccfdc2ad 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 sha256:55e31693954feaa2fb9bba61bd6b30f7f119ea0a7a3e4f18aefe5a72af9f351c -size 14455457 +oid sha256:6136c9676e800850762ab1152701a61382c92f61980841d108c11146a29bd9b1 +size 14530788 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 43665ea41ceef9f53ea9e7a1b9211f439a5bbed5..271a33c1d66a1d3bf89f2121cec04da72d703dca 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 sha256:0eecaefae3f2ed9046e67a913a9b6710d1bad97fd6fd0a48ab23d019b8139875 -size 73696 +oid sha256:99b0108722e9576f5d6a8d806819357999bb461cbc13f575db5c366bfbba325d +size 74414 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index f6347310470a4d161f97809f6ced7c468811d367..29dcbdf6d75b542965bbb2436ca4ffacf7d5a738 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": 0.690894, - "end_time": "2020-09-14T07:33:41.550115", + "duration": 3.758726, + "end_time": "2020-09-15T07:34:48.490913", "exception": false, - "start_time": "2020-09-14T07:33:40.859221", + "start_time": "2020-09-15T07:34:44.732187", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.022162, - "end_time": "2020-09-14T07:33:41.581043", + "duration": 0.029344, + "end_time": "2020-09-15T07:34:48.536522", "exception": false, - "start_time": "2020-09-14T07:33:41.558881", + "start_time": "2020-09-15T07:34:48.507178", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.01904, - "end_time": "2020-09-14T07:33:41.609092", + "duration": 0.033826, + "end_time": "2020-09-15T07:34:48.580975", "exception": false, - "start_time": "2020-09-14T07:33:41.590052", + "start_time": "2020-09-15T07:34:48.547149", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/oten11on/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/jsxpi5jx/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": 0.008436, - "end_time": "2020-09-14T07:33:41.626207", + "duration": 0.010067, + "end_time": "2020-09-15T07:34:48.605607", "exception": false, - "start_time": "2020-09-14T07:33:41.617771", + "start_time": "2020-09-15T07:34:48.595540", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.608125, - "end_time": "2020-09-14T07:33:42.241688", + "duration": 0.82058, + "end_time": "2020-09-15T07:34:49.433425", "exception": false, - "start_time": "2020-09-14T07:33:41.633563", + "start_time": "2020-09-15T07:34:48.612845", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.021432, - "end_time": "2020-09-14T07:33:42.272317", + "duration": 0.027218, + "end_time": "2020-09-15T07:34:49.474403", "exception": false, - "start_time": "2020-09-14T07:33:42.250885", + "start_time": "2020-09-15T07:34:49.447185", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.075222, - "end_time": "2020-09-14T07:33:42.356873", + "duration": 0.094807, + "end_time": "2020-09-15T07:34:49.578347", "exception": false, - "start_time": "2020-09-14T07:33:42.281651", + "start_time": "2020-09-15T07:34:49.483540", "status": "completed" }, "tags": [] @@ -263,10 +263,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008149, - "end_time": "2020-09-14T07:33:42.374696", + "duration": 0.007654, + "end_time": "2020-09-15T07:34:49.599510", "exception": false, - "start_time": "2020-09-14T07:33:42.366547", + "start_time": "2020-09-15T07:34:49.591856", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 24.154098, - "end_time": "2020-09-14T07:34:06.537107", + "duration": 24.500853, + "end_time": "2020-09-15T07:35:14.107416", "exception": false, - "start_time": "2020-09-14T07:33:42.383009", + "start_time": "2020-09-15T07:34:49.606563", "status": "completed" }, "tags": [] @@ -300,10 +300,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.038108, - "end_time": "2020-09-14T07:34:06.584415", + "duration": 0.052058, + "end_time": "2020-09-15T07:35:14.175197", "exception": false, - "start_time": "2020-09-14T07:34:06.546307", + "start_time": "2020-09-15T07:35:14.123139", "status": "completed" }, "tags": [] @@ -322,10 +322,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.561879, - "end_time": "2020-09-14T07:34:07.158624", + "duration": 0.515353, + "end_time": "2020-09-15T07:35:14.706843", "exception": false, - "start_time": "2020-09-14T07:34:06.596745", + "start_time": "2020-09-15T07:35:14.191490", "status": "completed" }, "tags": [] @@ -335,7 +335,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "10850 data points\n" + "10906 data points\n" ] }, { @@ -385,21 +385,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200913</td>\n", + " <td>20200914</td>\n", " <td>AK</td>\n", - " <td>7193.0</td>\n", - " <td>394020.0</td>\n", + " <td>7270.0</td>\n", + " <td>396143.0</td>\n", " <td>NaN</td>\n", - " <td>35.0</td>\n", + " <td>37.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>8.0</td>\n", " <td>...</td>\n", - " <td>401213</td>\n", + " <td>403413</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>f467ac0e20900a51da5e02bc4fb53260331b017c</td>\n", + " <td>c47ba6fdd7b2d5856a69b861c773b64ba0dfcb76</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -409,21 +409,21 @@ " </tr>\n", " <tr>\n", " <th>1</th>\n", - " <td>20200913</td>\n", + " <td>20200914</td>\n", " <td>AL</td>\n", - " <td>138755.0</td>\n", - " <td>884432.0</td>\n", + " <td>139459.0</td>\n", + " <td>888092.0</td>\n", " <td>NaN</td>\n", - " <td>790.0</td>\n", - " <td>15527.0</td>\n", + " <td>792.0</td>\n", + " <td>15756.0</td>\n", " <td>NaN</td>\n", - " <td>1610.0</td>\n", + " <td>1611.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>1023187</td>\n", - " <td>1</td>\n", - " <td>0</td>\n", - " <td>b84d1499d4e7dde4dca550ed4ab983ad7a428a9e</td>\n", + " <td>1027551</td>\n", + " <td>4</td>\n", + " <td>229</td>\n", + " <td>a852a907f21a1d2ec711d522217d2fa536b912e5</td>\n", " <td>0</td>\n", " <td>0</td>\n", " <td>0</td>\n", @@ -438,20 +438,20 @@ ], "text/plain": [ " date state positive negative pending hospitalizedCurrently \\\n", - "0 20200913 AK 7193.0 394020.0 NaN 35.0 \n", - "1 20200913 AL 138755.0 884432.0 NaN 790.0 \n", + "0 20200914 AK 7270.0 396143.0 NaN 37.0 \n", + "1 20200914 AL 139459.0 888092.0 NaN 792.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 15527.0 NaN 1610.0 \n", + "1 15756.0 NaN 1611.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 8.0 ... 401213 0 0 \n", - "1 NaN ... 1023187 1 0 \n", + "0 8.0 ... 403413 0 0 \n", + "1 NaN ... 1027551 4 229 \n", "\n", " hash commercialScore \\\n", - "0 f467ac0e20900a51da5e02bc4fb53260331b017c 0 \n", - "1 b84d1499d4e7dde4dca550ed4ab983ad7a428a9e 0 \n", + "0 c47ba6fdd7b2d5856a69b861c773b64ba0dfcb76 0 \n", + "1 a852a907f21a1d2ec711d522217d2fa536b912e5 0 \n", "\n", " negativeRegularScore negativeScore positiveScore score grade \n", "0 0 0 0 0 \n", @@ -491,18 +491,18 @@ "version": "3.7.3" }, "papermill": { - "duration": 27.930839, - "end_time": "2020-09-14T07:34:07.581638", + "duration": 32.417203, + "end_time": "2020-09-15T07:35:15.138903", "environment_variables": {}, "exception": null, - "input_path": "/tmp/oten11on/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/jsxpi5jx/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/oten11on/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/jsxpi5jx/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-09-14T07:33:39.650799", + "start_time": "2020-09-15T07:34:42.721700", "version": "1.1.0" } },