diff --git a/.renku/workflow/975bb67340ec472c964083369b804c54.cwl b/.renku/workflow/975bb67340ec472c964083369b804c54.cwl new file mode 100644 index 0000000000000000000000000000000000000000..9e05bcc7dee93e4b72bf898cff064da4fd2175eb --- /dev/null +++ b/.renku/workflow/975bb67340ec472c964083369b804c54.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_2: + outputSource: step_3/output_0 + streamable: false + type: File + output_3: + 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-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_1 + - output_0 + run: a17d560c41a54f5aa307ce5f3c5effe5_papermill.cwl diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index ed95224bec56527a5ce3fcbaefcae5aed95c6741..b44e8722f81b43a041abde8addddd63603f60a87 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 sha256:136defefdadf4133e1d20028db3d077e4ea2d141eee730b645374fe90e0e94d7 -size 18613184 +oid sha256:d13d414f947d832478298e57ee307398da1d358358bbbf245fb3f8ae3e767a34 +size 18767338 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index ab83d8c6df6d027658d625adcd14930894114872..4f88595696ce40a1934aff44b43afb016af493b1 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:f4ec8171dd99613eb1b81ecffa35355fe64aa13a460cc8b1a0dd3949e8933a1d -size 108213 +oid sha256:103a073640921a96c38ee84da94984a61e0bd11362acc39365a39cb70e828533 +size 109181 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index f8622d4e8e9d6d4f631b7b2535bfc7d25f4bb915..f4cc100edb928fe70a5f08b6b4698810a0c6ad74 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": 4.38731, - "end_time": "2020-11-02T07:40:01.538008", + "duration": 4.61991, + "end_time": "2020-11-04T07:36:05.231196", "exception": false, - "start_time": "2020-11-02T07:39:57.150698", + "start_time": "2020-11-04T07:36:00.611286", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.024656, - "end_time": "2020-11-02T07:40:01.578884", + "duration": 0.019678, + "end_time": "2020-11-04T07:36:05.263484", "exception": false, - "start_time": "2020-11-02T07:40:01.554228", + "start_time": "2020-11-04T07:36:05.243806", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.027968, - "end_time": "2020-11-02T07:40:01.620053", + "duration": 0.024073, + "end_time": "2020-11-04T07:36:05.295125", "exception": false, - "start_time": "2020-11-02T07:40:01.592085", + "start_time": "2020-11-04T07:36:05.271052", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/qj7q4luy/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/po7g_8x1/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": 0.008125, - "end_time": "2020-11-02T07:40:01.638870", + "duration": 0.008208, + "end_time": "2020-11-04T07:36:05.316133", "exception": false, - "start_time": "2020-11-02T07:40:01.630745", + "start_time": "2020-11-04T07:36:05.307925", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 1.010529, - "end_time": "2020-11-02T07:40:02.656587", + "duration": 0.691276, + "end_time": "2020-11-04T07:36:06.013731", "exception": false, - "start_time": "2020-11-02T07:40:01.646058", + "start_time": "2020-11-04T07:36:05.322455", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.025484, - "end_time": "2020-11-02T07:40:02.696195", + "duration": 0.025723, + "end_time": "2020-11-04T07:36:06.052061", "exception": false, - "start_time": "2020-11-02T07:40:02.670711", + "start_time": "2020-11-04T07:36:06.026338", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.081021, - "end_time": "2020-11-02T07:40:02.785478", + "duration": 0.073928, + "end_time": "2020-11-04T07:36:06.137689", "exception": false, - "start_time": "2020-11-02T07:40:02.704457", + "start_time": "2020-11-04T07:36:06.063761", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.008282, - "end_time": "2020-11-02T07:40:02.808385", + "duration": 0.009376, + "end_time": "2020-11-04T07:36:06.158189", "exception": false, - "start_time": "2020-11-02T07:40:02.800103", + "start_time": "2020-11-04T07:36:06.148813", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.486308, - "end_time": "2020-11-02T07:40:03.301771", + "duration": 0.461672, + "end_time": "2020-11-04T07:36:06.627415", "exception": false, - "start_time": "2020-11-02T07:40:02.815463", + "start_time": "2020-11-04T07:36:06.165743", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.06364, - "end_time": "2020-11-02T07:40:03.382133", + "duration": 0.048019, + "end_time": "2020-11-04T07:36:06.690160", "exception": false, - "start_time": "2020-11-02T07:40:03.318493", + "start_time": "2020-11-04T07:36:06.642141", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.707388, - "end_time": "2020-11-02T07:40:04.104186", + "duration": 0.5286, + "end_time": "2020-11-04T07:36:07.232531", "exception": false, - "start_time": "2020-11-02T07:40:03.396798", + "start_time": "2020-11-04T07:36:06.703931", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "13663 data points\n" + "13775 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201101</td>\n", + " <td>20201103</td>\n", " <td>AK</td>\n", - " <td>16705.0</td>\n", + " <td>17448.0</td>\n", " <td>NaN</td>\n", - " <td>587852.0</td>\n", + " <td>607307.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>604207.0</td>\n", - " <td>98.0</td>\n", + " <td>624755.0</td>\n", + " <td>97.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>604557</td>\n", - " <td>1</td>\n", + " <td>624755</td>\n", + " <td>0</td>\n", " <td>0</td>\n", - " <td>0fe81cb4b3a17588cfde8bbd309c3f0f2e786fd2</td>\n", + " <td>e4b7b20d6a7724e14ce53e37e1d543c6bf1176c9</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>20201101</td>\n", + " <td>20201103</td>\n", " <td>AL</td>\n", - " <td>193985.0</td>\n", - " <td>28746.0</td>\n", - " <td>1187341.0</td>\n", + " <td>195929.0</td>\n", + " <td>29302.0</td>\n", + " <td>1196422.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1352580.0</td>\n", - " <td>967.0</td>\n", - " <td>20450.0</td>\n", + " <td>1363049.0</td>\n", + " <td>1022.0</td>\n", + " <td>20901.0</td>\n", " <td>...</td>\n", - " <td>1381326</td>\n", - " <td>6</td>\n", - " <td>0</td>\n", - " <td>0bd0f2572848cf6f46a651c076d32afceb8dc6a1</td>\n", + " <td>1392351</td>\n", + " <td>14</td>\n", + " <td>451</td>\n", + " <td>f6dd94c1b534c3e4a74e78d1859fcc12ae1ef103</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 20201101 AK 16705.0 NaN 587852.0 NaN \n", - "1 20201101 AL 193985.0 28746.0 1187341.0 NaN \n", + "0 20201103 AK 17448.0 NaN 607307.0 NaN \n", + "1 20201103 AL 195929.0 29302.0 1196422.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 604207.0 98.0 \n", - "1 totalTestsViral 1352580.0 967.0 \n", + "0 totalTestsViral 624755.0 97.0 \n", + "1 totalTestsViral 1363049.0 1022.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 604557 1 0 \n", - "1 20450.0 ... 1381326 6 0 \n", + "0 NaN ... 624755 0 0 \n", + "1 20901.0 ... 1392351 14 451 \n", "\n", " hash commercialScore \\\n", - "0 0fe81cb4b3a17588cfde8bbd309c3f0f2e786fd2 0 \n", - "1 0bd0f2572848cf6f46a651c076d32afceb8dc6a1 0 \n", + "0 e4b7b20d6a7724e14ce53e37e1d543c6bf1176c9 0 \n", + "1 f6dd94c1b534c3e4a74e78d1859fcc12ae1ef103 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": 9.174658, - "end_time": "2020-11-02T07:40:04.530883", + "duration": 8.604898, + "end_time": "2020-11-04T07:36:07.556382", "environment_variables": {}, "exception": null, - "input_path": "/tmp/qj7q4luy/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/po7g_8x1/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/qj7q4luy/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/po7g_8x1/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-02T07:39:55.356225", + "start_time": "2020-11-04T07:35:58.951484", "version": "1.1.0" } },