{ "id": 225517, "title": "Problem with testing data", "url": "/competitions/indoor-location-navigation/discussion/225517", "author_name": "", "post_date": "2021-03-12T16:45:52.878821100Z", "votes": 2, "comment_count": 2, "views": 0, "content": "

In rows which are about WiFi connection there is Timestamp (in the first column) and Last seen (in the last column).

\n

I would suppose the second value to be greater or equal to the first value. In the training data it holds (as far as I have checked). But in the testing data Timestamp starts at 0, and the condition doesn't hold.

\n

Example:
\n0000000002139 TYPE_WIFI da39a3ee5e6b4b0d3255bfef95601890afd80709 7a626b941b8fa40da1f730c79b169bb2ab4506ae -46 5180 1574146843910

\n

0000000002139 TYPE_WIFI da39a3ee5e6b4b0d3255bfef95601890afd80709 4ad3076e3d2907f278578ee8165ac32700d6de74 -46 5180 1574146843911

\n

0000000002139 TYPE_WIFI 8f7d5669fb8464cd2eef926f246adaac8050b848 ebda195d96124ca68eaa98d32d97e9f4577289d0 -47 5180 1574146843910

\n

As they told in the video, dealing with this gap might be crucial.
\nSo does anybody know why there is such inconsistency in testing data?

", "messages": [ { "id": "1235991", "postDate": "03/12/2021 16:45:52", "content": "

In rows which are about WiFi connection there is Timestamp (in the first column) and Last seen (in the last column).

\n

I would suppose the second value to be greater or equal to the first value. In the training data it holds (as far as I have checked). But in the testing data Timestamp starts at 0, and the condition doesn't hold.

\n

Example:
\n0000000002139 TYPE_WIFI da39a3ee5e6b4b0d3255bfef95601890afd80709 7a626b941b8fa40da1f730c79b169bb2ab4506ae -46 5180 1574146843910

\n

0000000002139 TYPE_WIFI da39a3ee5e6b4b0d3255bfef95601890afd80709 4ad3076e3d2907f278578ee8165ac32700d6de74 -46 5180 1574146843911

\n

0000000002139 TYPE_WIFI 8f7d5669fb8464cd2eef926f246adaac8050b848 ebda195d96124ca68eaa98d32d97e9f4577289d0 -47 5180 1574146843910

\n

As they told in the video, dealing with this gap might be crucial.
\nSo does anybody know why there is such inconsistency in testing data?

", "rawMarkdown": "In rows which are about WiFi connection there is ***Timestamp*** (in the first column) and ***Last seen*** (in the last column). \n\nI would suppose the second value to be greater or equal to the first value. In the training data it holds (as far as I have checked). But in the testing data *Timestamp* starts at 0, and the condition doesn't hold.\n\nExample:\n0000000002139\tTYPE_WIFI\tda39a3ee5e6b4b0d3255bfef95601890afd80709\t7a626b941b8fa40da1f730c79b169bb2ab4506ae\t-46\t5180\t1574146843910\n\n0000000002139\tTYPE_WIFI\tda39a3ee5e6b4b0d3255bfef95601890afd80709\t4ad3076e3d2907f278578ee8165ac32700d6de74\t-46\t5180\t1574146843911\n\n0000000002139\tTYPE_WIFI\t8f7d5669fb8464cd2eef926f246adaac8050b848\tebda195d96124ca68eaa98d32d97e9f4577289d0\t-47\t5180\t1574146843910\n\nAs they told in the video, dealing with this gap might be crucial.\nSo does anybody know why there is such inconsistency in testing data?", "votes": null }, { "id": "1236049", "postDate": "03/12/2021 17:50:53", "content": "

Please look at this thread. I'm making a notebook to show how to fix it. I'll publish it when it's done.

", "rawMarkdown": "Please look at this [thread](https://www.kaggle.com/c/indoor-location-navigation/discussion/218074). I'm making a notebook to show how to fix it. I'll publish it when it's done.", "votes": null }, { "id": "1236085", "postDate": "03/12/2021 18:39:17", "content": "

here it is: https://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask

", "rawMarkdown": "here it is: https://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask", "votes": null } ], "comments": [ { "id": 1236049, "author_name": "jiweiliu", "author_url": "", "post_date": "03/12/2021 17:50:53", "content": "

Please look at this thread. I'm making a notebook to show how to fix it. I'll publish it when it's done.

", "votes": null, "replies": [] }, { "id": 1236085, "author_name": "jiweiliu", "author_url": "", "post_date": "03/12/2021 18:39:17", "content": "

here it is: https://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1235991": "In rows which are about WiFi connection there is ***Timestamp*** (in the first column) and ***Last seen*** (in the last column). \n\nI would suppose the second value to be greater or equal to the first value. In the training data it holds (as far as I have checked). But in the testing data *Timestamp* starts at 0, and the condition doesn't hold.\n\nExample:\n0000000002139\tTYPE_WIFI\tda39a3ee5e6b4b0d3255bfef95601890afd80709\t7a626b941b8fa40da1f730c79b169bb2ab4506ae\t-46\t5180\t1574146843910\n\n0000000002139\tTYPE_WIFI\tda39a3ee5e6b4b0d3255bfef95601890afd80709\t4ad3076e3d2907f278578ee8165ac32700d6de74\t-46\t5180\t1574146843911\n\n0000000002139\tTYPE_WIFI\t8f7d5669fb8464cd2eef926f246adaac8050b848\tebda195d96124ca68eaa98d32d97e9f4577289d0\t-47\t5180\t1574146843910\n\nAs they told in the video, dealing with this gap might be crucial.\nSo does anybody know why there is such inconsistency in testing data?", "1236049": "Please look at this [thread](https://www.kaggle.com/c/indoor-location-navigation/discussion/218074). I'm making a notebook to show how to fix it. I'll publish it when it's done.", "1236085": "here it is: https://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask" }, "source": "meta" }