diff --git a/.renku/workflow/996b6b86b965481f96d76a6f19518a32.cwl b/.renku/workflow/996b6b86b965481f96d76a6f19518a32.cwl new file mode 100644 index 0000000000000000000000000000000000000000..a70cab1b0894b45fdb6210726544b051faead6ca --- /dev/null +++ b/.renku/workflow/996b6b86b965481f96d76a6f19518a32.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-metadata.json + streamable: false + type: string + input_6: + default: states-daily.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_1 + - output_0 + 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-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_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-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: [] diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json index 0ca3ce03c1a9b10f63a02f59720aa50c549fbed9..d71d510131864e376d754a41cf6833afb98dcdaf 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:a624d3f17242e07532ae74e0a91bceb2562e10c5d51d52aaae2064a6c43a4a0a -size 19306923 +oid sha256:babd48263f9decd6af9d82c13cf75de23e3938855b29d61277071026e9c0c360 +size 19384110 diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json index 0d51c71206c71c8455bcb307b24f932aa353b283..6bc4c450ed665eff92045cd53e4ffa8525d7b77d 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 sha256:4820e5e0b14cdf83f2adbca3b264c719c612711e940e0e53cfb18b795d99873c -size 116902 +oid sha256:529c259e9dfe1bc1256570ade6588b0e62b2648db3f22bdb8d8af6af75bbb5d6 +size 118467 diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb index 9a5dc3172c32606f309127295a44cac6525e7fe9..be3556398d8bf4e942d410cb014ae1ee3c29ce0d 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": 3.486428, - "end_time": "2020-11-11T07:37:20.490952", + "duration": 5.140415, + "end_time": "2020-11-12T07:37:44.046695", "exception": false, - "start_time": "2020-11-11T07:37:17.004524", + "start_time": "2020-11-12T07:37:38.906280", "status": "completed" }, "tags": [] @@ -25,10 +25,10 @@ "execution_count": 2, "metadata": { "papermill": { - "duration": 0.021188, - "end_time": "2020-11-11T07:37:20.525455", + "duration": 0.021454, + "end_time": "2020-11-12T07:37:44.082314", "exception": false, - "start_time": "2020-11-11T07:37:20.504267", + "start_time": "2020-11-12T07:37:44.060860", "status": "completed" }, "tags": [ @@ -46,10 +46,10 @@ "execution_count": 3, "metadata": { "papermill": { - "duration": 0.020446, - "end_time": "2020-11-11T07:37:20.554599", + "duration": 0.018562, + "end_time": "2020-11-12T07:37:44.108496", "exception": false, - "start_time": "2020-11-11T07:37:20.534153", + "start_time": "2020-11-12T07:37:44.089934", "status": "completed" }, "tags": [ @@ -59,7 +59,7 @@ "outputs": [], "source": [ "# Parameters\n", - "PAPERMILL_INPUT_PATH = \"/tmp/jxtcd406/notebooks/process/download-covidtracking-data.ipynb\"\n", + "PAPERMILL_INPUT_PATH = \"/tmp/g2e0cwhg/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": 0.007527, - "end_time": "2020-11-11T07:37:20.570224", + "duration": 0.005836, + "end_time": "2020-11-12T07:37:44.120706", "exception": false, - "start_time": "2020-11-11T07:37:20.562697", + "start_time": "2020-11-12T07:37:44.114870", "status": "completed" }, "tags": [] @@ -87,10 +87,10 @@ "execution_count": 4, "metadata": { "papermill": { - "duration": 0.840977, - "end_time": "2020-11-11T07:37:21.416769", + "duration": 0.385961, + "end_time": "2020-11-12T07:37:44.511193", "exception": false, - "start_time": "2020-11-11T07:37:20.575792", + "start_time": "2020-11-12T07:37:44.125232", "status": "completed" }, "tags": [] @@ -107,10 +107,10 @@ "execution_count": 5, "metadata": { "papermill": { - "duration": 0.02556, - "end_time": "2020-11-11T07:37:21.456405", + "duration": 0.022275, + "end_time": "2020-11-12T07:37:44.547570", "exception": false, - "start_time": "2020-11-11T07:37:21.430845", + "start_time": "2020-11-12T07:37:44.525295", "status": "completed" }, "tags": [] @@ -129,10 +129,10 @@ "execution_count": 6, "metadata": { "papermill": { - "duration": 0.072094, - "end_time": "2020-11-11T07:37:21.541562", + "duration": 0.064783, + "end_time": "2020-11-12T07:37:44.619971", "exception": false, - "start_time": "2020-11-11T07:37:21.469468", + "start_time": "2020-11-12T07:37:44.555188", "status": "completed" }, "tags": [] @@ -280,10 +280,10 @@ "cell_type": "markdown", "metadata": { "papermill": { - "duration": 0.007925, - "end_time": "2020-11-11T07:37:21.561857", + "duration": 0.007389, + "end_time": "2020-11-12T07:37:44.639765", "exception": false, - "start_time": "2020-11-11T07:37:21.553932", + "start_time": "2020-11-12T07:37:44.632376", "status": "completed" }, "tags": [] @@ -297,10 +297,10 @@ "execution_count": 7, "metadata": { "papermill": { - "duration": 0.443584, - "end_time": "2020-11-11T07:37:22.010827", + "duration": 0.455541, + "end_time": "2020-11-12T07:37:45.101712", "exception": false, - "start_time": "2020-11-11T07:37:21.567243", + "start_time": "2020-11-12T07:37:44.646171", "status": "completed" }, "tags": [] @@ -317,10 +317,10 @@ "execution_count": 8, "metadata": { "papermill": { - "duration": 0.047709, - "end_time": "2020-11-11T07:37:22.074585", + "duration": 0.026651, + "end_time": "2020-11-12T07:37:45.134927", "exception": false, - "start_time": "2020-11-11T07:37:22.026876", + "start_time": "2020-11-12T07:37:45.108276", "status": "completed" }, "tags": [] @@ -339,10 +339,10 @@ "execution_count": 9, "metadata": { "papermill": { - "duration": 0.536519, - "end_time": "2020-11-11T07:37:22.626284", + "duration": 0.508568, + "end_time": "2020-11-12T07:37:45.648467", "exception": false, - "start_time": "2020-11-11T07:37:22.089765", + "start_time": "2020-11-12T07:37:45.139899", "status": "completed" }, "tags": [] @@ -352,7 +352,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "14167 data points\n" + "14223 data points\n" ] }, { @@ -402,21 +402,21 @@ " <tbody>\n", " <tr>\n", " <th>0</th>\n", - " <td>20201110</td>\n", + " <td>20201111</td>\n", " <td>AK</td>\n", - " <td>20832.0</td>\n", + " <td>21331.0</td>\n", " <td>NaN</td>\n", - " <td>807308.0</td>\n", + " <td>813259.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>828140.0</td>\n", - " <td>120.0</td>\n", + " <td>834590.0</td>\n", + " <td>117.0</td>\n", " <td>NaN</td>\n", " <td>...</td>\n", - " <td>828140</td>\n", - " <td>8</td>\n", + " <td>834590</td>\n", + " <td>4</td>\n", " <td>0</td>\n", - " <td>46e28ec7ada24608af43b6ea894722e776bc3c2c</td>\n", + " <td>54c6a77d7d1f67807c0d8c1ff07f4b3ab0132140</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>20201110</td>\n", + " <td>20201111</td>\n", " <td>AL</td>\n", - " <td>206567.0</td>\n", - " <td>31580.0</td>\n", - " <td>1231152.0</td>\n", + " <td>208637.0</td>\n", + " <td>32282.0</td>\n", + " <td>1245235.0</td>\n", " <td>NaN</td>\n", " <td>totalTestsViral</td>\n", - " <td>1406139.0</td>\n", - " <td>1206.0</td>\n", + " <td>1421590.0</td>\n", + " <td>1210.0</td>\n", " <td>21886.0</td>\n", " <td>...</td>\n", - " <td>1437719</td>\n", - " <td>36</td>\n", - " <td>592</td>\n", - " <td>9b5e8f2e29cb12f58314accdd17ba4c4a8e15a65</td>\n", + " <td>1453872</td>\n", + " <td>81</td>\n", + " <td>0</td>\n", + " <td>a7a21b2a579a0ef9c3c9fdce8e950c09220d5985</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 20201110 AK 20832.0 NaN 807308.0 NaN \n", - "1 20201110 AL 206567.0 31580.0 1231152.0 NaN \n", + "0 20201111 AK 21331.0 NaN 813259.0 NaN \n", + "1 20201111 AL 208637.0 32282.0 1245235.0 NaN \n", "\n", " totalTestResultsSource totalTestResults hospitalizedCurrently \\\n", - "0 totalTestsViral 828140.0 120.0 \n", - "1 totalTestsViral 1406139.0 1206.0 \n", + "0 totalTestsViral 834590.0 117.0 \n", + "1 totalTestsViral 1421590.0 1210.0 \n", "\n", " hospitalizedCumulative ... posNeg deathIncrease hospitalizedIncrease \\\n", - "0 NaN ... 828140 8 0 \n", - "1 21886.0 ... 1437719 36 592 \n", + "0 NaN ... 834590 4 0 \n", + "1 21886.0 ... 1453872 81 0 \n", "\n", " hash commercialScore \\\n", - "0 46e28ec7ada24608af43b6ea894722e776bc3c2c 0 \n", - "1 9b5e8f2e29cb12f58314accdd17ba4c4a8e15a65 0 \n", + "0 54c6a77d7d1f67807c0d8c1ff07f4b3ab0132140 0 \n", + "1 a7a21b2a579a0ef9c3c9fdce8e950c09220d5985 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": 7.908161, - "end_time": "2020-11-11T07:37:22.952647", + "duration": 8.905717, + "end_time": "2020-11-12T07:37:45.963024", "environment_variables": {}, "exception": null, - "input_path": "/tmp/jxtcd406/notebooks/process/download-covidtracking-data.ipynb", + "input_path": "/tmp/g2e0cwhg/notebooks/process/download-covidtracking-data.ipynb", "output_path": "runs/download-covidtracking-data.runs.ipynb", "parameters": { - "PAPERMILL_INPUT_PATH": "/tmp/jxtcd406/notebooks/process/download-covidtracking-data.ipynb", + "PAPERMILL_INPUT_PATH": "/tmp/g2e0cwhg/notebooks/process/download-covidtracking-data.ipynb", "PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb", "out_folder": "data/covidtracking" }, - "start_time": "2020-11-11T07:37:15.044486", + "start_time": "2020-11-12T07:37:37.057307", "version": "1.1.0" } },