diff --git a/.renku/workflow/ec866df5f2c0429ca86f04829489e0b0.cwl b/.renku/workflow/ec866df5f2c0429ca86f04829489e0b0.cwl new file mode 100644 index 0000000000000000000000000000000000000000..4dd78826e1abf9a4aee255083a983b5e736fe094 --- /dev/null +++ b/.renku/workflow/ec866df5f2c0429ca86f04829489e0b0.cwl @@ -0,0 +1,120 @@ +class: Workflow +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: states-metadata.json + streamable: false + type: string + input_2: + default: states-daily.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_0: + outputSource: step_3/output_1 + streamable: false + type: Directory + output_3: + outputSource: step_3/output_0 + streamable: false + type: File +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-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 + 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-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 + - *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 6dd10404fcaf8e8605e0ee12aef3000125ea7a40..b24a82eabb5b0aa649b2221efebad0a4006c571f 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 sha256:1db64bf05ba50bd7a87d3ac5ec39cbdffabac74019e66066ad6d869f7a961ea0 -size 6818612 +oid sha256:07d8fb27d4370ea29687e03981bd732cfbf647dbd1aed89256d1a4e489ab49c7 +size 6871224 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 3fc6339f8e675720db62256fd26449bf198b286e..04e820843efef880c4eb84cffea4a56cc6c06423 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 sha256:6d6512ed1bdfcac4a24fa5fccbdf9b06e76d9187eb6a707766831fc7589382ca -size 48897 +oid sha256:45a2572bd641eeac8a4bda332023f0fc319b6507dc49b3fb819bc85897f9c65e +size 49640 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 8caf4186af28774fd4aca1c7b4a8832cc8a40bcd..4aae90d032e087a4d8fa5b84645f8af43a496307 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": 2.265336, - "end_time": "2020-07-15T07:31:11.645868", + "duration": 5.211328, + "end_time": "2020-07-16T07:36:32.349184", "exception": false, - "start_time": "2020-07-15T07:31:09.380532", + "start_time": "2020-07-16T07:36:27.137856", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.021481, - "end_time": "2020-07-15T07:31:11.681445", + "duration": 0.020238, + "end_time": "2020-07-16T07:36:32.380757", "exception": false, - "start_time": "2020-07-15T07:31:11.659964", + "start_time": "2020-07-16T07:36:32.360519", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.025026, - "end_time": "2020-07-15T07:31:11.715032", + "duration": 0.020212, + "end_time": "2020-07-16T07:36:32.412094", "exception": false, - "start_time": "2020-07-15T07:31:11.690006", + "start_time": "2020-07-16T07:36:32.391882", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/5hekzgtt/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/z5hwmmrn/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": 0.007207, - "end_time": "2020-07-15T07:31:11.734576", + "duration": 0.009151, + "end_time": "2020-07-16T07:36:32.432210", "exception": false, - "start_time": "2020-07-15T07:31:11.727369", + "start_time": "2020-07-16T07:36:32.423059", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.094813, - "end_time": "2020-07-15T07:31:11.835983", + "duration": 0.193284, + "end_time": "2020-07-16T07:36:32.634328", "exception": false, - "start_time": "2020-07-15T07:31:11.741170", + "start_time": "2020-07-16T07:36:32.441044", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.023474, - "end_time": "2020-07-15T07:31:11.867961", + "duration": 0.021824, + "end_time": "2020-07-16T07:36:32.667175", "exception": false, - "start_time": "2020-07-15T07:31:11.844487", + "start_time": "2020-07-16T07:36:32.645351", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.078157, - "end_time": "2020-07-15T07:31:11.953010", + "duration": 0.071576, + "end_time": "2020-07-16T07:36:32.749621", "exception": false, - "start_time": "2020-07-15T07:31:11.874853", + "start_time": "2020-07-16T07:36:32.678045", "status": "completed" }, "tags": [] @@ -249,10 +249,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.009274, - "end_time": "2020-07-15T07:31:11.974426", + "duration": 0.008642, + "end_time": "2020-07-16T07:36:32.769476", "exception": false, - "start_time": "2020-07-15T07:31:11.965152", + "start_time": "2020-07-16T07:36:32.760834", "status": "completed" }, "tags": [] @@ -266,10 +266,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.260704, - "end_time": "2020-07-15T07:31:12.243305", + "duration": 0.308814, + "end_time": "2020-07-16T07:36:33.087074", "exception": false, - "start_time": "2020-07-15T07:31:11.982601", + "start_time": "2020-07-16T07:36:32.778260", "status": "completed" }, "tags": [] @@ -286,10 +286,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.037799, - "end_time": "2020-07-15T07:31:12.298355", + "duration": 0.0299, + "end_time": "2020-07-16T07:36:33.129622", "exception": false, - "start_time": "2020-07-15T07:31:12.260556", + "start_time": "2020-07-16T07:36:33.099722", "status": "completed" }, "tags": [] @@ -308,10 +308,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.29273, - "end_time": "2020-07-15T07:31:12.605093", + "duration": 0.274984, + "end_time": "2020-07-16T07:36:33.418558", "exception": false, - "start_time": "2020-07-15T07:31:12.312363", + "start_time": "2020-07-16T07:36:33.143574", "status": "completed" }, "tags": [] @@ -321,7 +321,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "7353 data points\n" + "7409 data points\n" ] }, { @@ -371,21 +371,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20200714</td>\n", + " <td>20200715</td>\n", " <td>AK</td>\n", - " <td>1579.0</td>\n", - " <td>147894.0</td>\n", + " <td>1631.0</td>\n", + " <td>154462.0</td>\n", " <td>NaN</td>\n", - " <td>25.0</td>\n", + " <td>32.0</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", " <td>NaN</td>\n", - " <td>1.0</td>\n", + " <td>0.0</td>\n", " <td>...</td>\n", - " <td>149473</td>\n", + " <td>156093</td>\n", " <td>0</td>\n", " <td>0</td>\n", - " <td>14ec620d73185257a210b3f634cca1a8024dabe0</td>\n", + " <td>94f51a21cc1532dde7387df47013f567fb56d967</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>20200714</td>\n", + " <td>20200715</td>\n", " <td>AL</td>\n", - " <td>57255.0</td>\n", - " <td>471834.0</td>\n", + " <td>59067.0</td>\n", + " <td>481982.0</td>\n", " <td>NaN</td>\n", - " <td>1353.0</td>\n", - " <td>7123.0</td>\n", + " <td>1362.0</td>\n", + " <td>7291.0</td>\n", " <td>NaN</td>\n", - " <td>931.0</td>\n", + " <td>941.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>529089</td>\n", - " <td>40</td>\n", - " <td>378</td>\n", - " <td>820cd0cab43d1d4994f54d9f80d70d3ce3257dbe</td>\n", + " <td>541049</td>\n", + " <td>47</td>\n", + " <td>168</td>\n", + " <td>115a76b81b0c3c59d334b5de8180c2e10899dddc</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 20200714 AK 1579.0 147894.0 NaN 25.0 \n", - "1 20200714 AL 57255.0 471834.0 NaN 1353.0 \n", + "0 20200715 AK 1631.0 154462.0 NaN 32.0 \n", + "1 20200715 AL 59067.0 481982.0 NaN 1362.0 \n", "\n", " hospitalizedCumulative inIcuCurrently inIcuCumulative \\\n", "0 NaN NaN NaN \n", - "1 7123.0 NaN 931.0 \n", + "1 7291.0 NaN 941.0 \n", "\n", " onVentilatorCurrently ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 1.0 ... 149473 0 0 \n", - "1 NaN ... 529089 40 378 \n", + "0 0.0 ... 156093 0 0 \n", + "1 NaN ... 541049 47 168 \n", "\n", " hash commercialScore \\\n", - "0 14ec620d73185257a210b3f634cca1a8024dabe0 0 \n", - "1 820cd0cab43d1d4994f54d9f80d70d3ce3257dbe 0 \n", + "0 94f51a21cc1532dde7387df47013f567fb56d967 0 \n", + "1 115a76b81b0c3c59d334b5de8180c2e10899dddc 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": 5.225562, - "end_time": "2020-07-15T07:31:12.927967", + "duration": 8.748295, + "end_time": "2020-07-16T07:36:33.741465", "environment_variables": {}, "exception": null, - "input_path": "/tmp/5hekzgtt/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/z5hwmmrn/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/5hekzgtt/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/z5hwmmrn/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-07-15T07:31:07.702405", + "start_time": "2020-07-16T07:36:24.993170", "version": "1.1.0" } },