![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3081183%2Fd5c41e57674d0750c3a0b9c944c43d57%2FCapturar.JPG?generation=1698506187366053&alt=media\")
Hi all,
\nI have trying to find a filter to remove region that the device is not being used in the series, a lot of articles filter it by rolling standard deviation, i made this notebook to see that pattern in data, after some test I made a combination of metric the filter this. When the metric is less than 21 i decide to remove this part.
\nSeries with some null Events:
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3081183%2Fe243101c19148463ed6a3ff4852de672%2FCapturar.JPG?generation=1698506415699292&alt=media\")
Obs: The filter is a bit conservative and does not remove it every time the device is not being used, but it is useful when the device is not used for several days in a row
", "messages": [ { "id": "2502934", "postDate": "10/28/2023 15:16:38", "content": "Hi all,
\nI have trying to find a filter to remove region that the device is not being used in the series, a lot of articles filter it by rolling standard deviation, i made this notebook to see that pattern in data, after some test I made a combination of metric the filter this. When the metric is less than 21 i decide to remove this part.
\nSeries with some null Events:
\n
Obs: The filter is a bit conservative and does not remove it every time the device is not being used, but it is useful when the device is not used for several days in a row
", "rawMarkdown": "Hi all,\n\nI have trying to find a filter to remove region that the device is not being used in the series, a lot of articles filter it by rolling standard deviation, i made this [notebook](https://www.kaggle.com/code/renatoreggiani/zzzs-viz-for-not-wear-events/notebook?scriptVersionId=148354374) to see that pattern in data, after some test I made a combination of metric the filter this. When the metric is less than 21 i decide to remove this part.\n\nSeries with some null Events:\n\n \n\n\n\n\nObs: The filter is a bit conservative and does not remove it every time the device is not being used, but it is useful when the device is not used for several days in a row", "votes": null }, { "id": "2508317", "postDate": "11/01/2023 15:51:10", "content": "@renatoreggiani : thanks for sharing the interesting observations, Renato!
", "rawMarkdown": "renatoreggiani : thanks for sharing the interesting observations, Renato!", "votes": null } ], "comments": [ { "id": 2508317, "author_name": "gvyshnya", "author_url": "", "post_date": "11/01/2023 15:51:10", "content": "@renatoreggiani : thanks for sharing the interesting observations, Renato!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2502934": "Hi all,\n\nI have trying to find a filter to remove region that the device is not being used in the series, a lot of articles filter it by rolling standard deviation, i made this [notebook](https://www.kaggle.com/code/renatoreggiani/zzzs-viz-for-not-wear-events/notebook?scriptVersionId=148354374) to see that pattern in data, after some test I made a combination of metric the filter this. When the metric is less than 21 i decide to remove this part.\n\nSeries with some null Events:\n\n \n\n\n\n\nObs: The filter is a bit conservative and does not remove it every time the device is not being used, but it is useful when the device is not used for several days in a row", "2508317": "renatoreggiani : thanks for sharing the interesting observations, Renato!" }, "source": "meta" }