{ "id": 230153, "title": "Uses of public notebooks", "url": "/competitions/indoor-location-navigation/discussion/230153", "author_name": "", "post_date": "2021-04-02T08:46:36.242176200Z", "votes": 35, "comment_count": 2, "views": 0, "content": "

I would like to mention a number of public notebooks that we have used.
\nThanks to everyone who has published their notebooks publicly.

\n

Fix the floor prediction

\n

We used the following great notebook for \"Floor prediction\". You can also try the ['floor'] column of the results of this notebook.

\n

https://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model

\n

Ensembling

\n

We used a lot of public notebooks for \"Ensembling\" that scored very well. The addresses of some of them are as follows:

\n

https://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing

\n

https://www.kaggle.com/therocket290/lstm-unified-wi-fi-training-x-and-y-with-floor

\n

https://www.kaggle.com/kokitanisaka/lstm-by-keras-with-unified-wi-fi-feats

\n

https://www.kaggle.com/oxzplvifi/indoor-gbm-postprocessing-xy-prediction

\n

https://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model

\n

https://www.kaggle.com/hiro5299834/wifi-features-with-lightgbm-kfold

\n

https://www.kaggle.com/ebinan92/time-series-rnn-xy-prediction

\n

We released the following notebook to do \"Ensembling\". We used different methods in this notebook. For example, to eliminate bad answers in some stations, we used the \"decision rectangle\".

\n

https://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method

\n

Cost Minimization

\n

A magical notebook has been released that shows all the results can be improved. The address of this notebook is as follows:

\n

https://www.kaggle.com/saitodevel01/indoor-post-processing-by-cost-minimization

\n

You can use this notebook before or after \"Ensembling\". Using this notebook, we also improved the score of each notebook before \"Ensembling\" and at the same time equalized the ['floor'] column for all of them. Our notebook address is as follows:

\n

https://www.kaggle.com/mehrankazeminia/1-3-indoor-navigation-cost-minimization

\n

Push to hallway

\n

Several notebooks have been published for this purpose. We used the excellent notebook below, which is called \"Snap to Grid\". We use this method only after \"Ensembling\". But if you use this method for each notebook before \"Ensembling\", several errors will accumulate and it is not clear that you can get a good result.

\n

https://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing

\n

The author explains in this notebook that you can try numbers from 3 to 8 to determine the value for \"Threshold\". Of course, in our notebook, whose address is as follows, we used two values for \"Threshold\" at the same time to get a better result.

\n

https://www.kaggle.com/mehrankazeminia/3-3-g6-indoor-navigation-snap-to-grid

\n

In the following notebook, a better way to determine the value of \"Threshold\" is done. The more numbers you try, the more accurate the number for the \"Threshold\" value. Of course, for different results, the value of \"Threshold\" also changes.

\n

https://www.kaggle.com/dragonzhang/3-3-g6-indoor-navigation-snap-to-grid

\n

Fix the timestamps

\n

In the excellent notebook below, the problem is described in the [\"site_path_timestamp\"] column and suggestions are made:

\n

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

\n

Then another notebook was released which is a development of the previous notebook and has excellent details.

\n

https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage

\n

We also used the above notebook to improve our results.

\n

Good Luck

", "messages": [ { "id": "1260595", "postDate": "04/02/2021 08:46:36", "content": "

I would like to mention a number of public notebooks that we have used.
\nThanks to everyone who has published their notebooks publicly.

\n

Fix the floor prediction

\n

We used the following great notebook for \"Floor prediction\". You can also try the ['floor'] column of the results of this notebook.

\n

https://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model

\n

Ensembling

\n

We used a lot of public notebooks for \"Ensembling\" that scored very well. The addresses of some of them are as follows:

\n

https://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing

\n

https://www.kaggle.com/therocket290/lstm-unified-wi-fi-training-x-and-y-with-floor

\n

https://www.kaggle.com/kokitanisaka/lstm-by-keras-with-unified-wi-fi-feats

\n

https://www.kaggle.com/oxzplvifi/indoor-gbm-postprocessing-xy-prediction

\n

https://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model

\n

https://www.kaggle.com/hiro5299834/wifi-features-with-lightgbm-kfold

\n

https://www.kaggle.com/ebinan92/time-series-rnn-xy-prediction

\n

We released the following notebook to do \"Ensembling\". We used different methods in this notebook. For example, to eliminate bad answers in some stations, we used the \"decision rectangle\".

\n

https://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method

\n

Cost Minimization

\n

A magical notebook has been released that shows all the results can be improved. The address of this notebook is as follows:

\n

https://www.kaggle.com/saitodevel01/indoor-post-processing-by-cost-minimization

\n

You can use this notebook before or after \"Ensembling\". Using this notebook, we also improved the score of each notebook before \"Ensembling\" and at the same time equalized the ['floor'] column for all of them. Our notebook address is as follows:

\n

https://www.kaggle.com/mehrankazeminia/1-3-indoor-navigation-cost-minimization

\n

Push to hallway

\n

Several notebooks have been published for this purpose. We used the excellent notebook below, which is called \"Snap to Grid\". We use this method only after \"Ensembling\". But if you use this method for each notebook before \"Ensembling\", several errors will accumulate and it is not clear that you can get a good result.

\n

https://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing

\n

The author explains in this notebook that you can try numbers from 3 to 8 to determine the value for \"Threshold\". Of course, in our notebook, whose address is as follows, we used two values for \"Threshold\" at the same time to get a better result.

\n

https://www.kaggle.com/mehrankazeminia/3-3-g6-indoor-navigation-snap-to-grid

\n

In the following notebook, a better way to determine the value of \"Threshold\" is done. The more numbers you try, the more accurate the number for the \"Threshold\" value. Of course, for different results, the value of \"Threshold\" also changes.

\n

https://www.kaggle.com/dragonzhang/3-3-g6-indoor-navigation-snap-to-grid

\n

Fix the timestamps

\n

In the excellent notebook below, the problem is described in the [\"site_path_timestamp\"] column and suggestions are made:

\n

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

\n

Then another notebook was released which is a development of the previous notebook and has excellent details.

\n

https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage

\n

We also used the above notebook to improve our results.

\n

Good Luck

", "rawMarkdown": "I would like to mention a number of public notebooks that we have used. \nThanks to everyone who has published their notebooks publicly.\n\n****Fix the floor prediction****\n\nWe used the following great notebook for \"Floor prediction\". You can also try the ['floor'] column of the results of this notebook.\n\nhttps://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model\n\n****Ensembling****\n\nWe used a lot of public notebooks for \"Ensembling\" that scored very well. The addresses of some of them are as follows:\n\nhttps://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing\n\nhttps://www.kaggle.com/therocket290/lstm-unified-wi-fi-training-x-and-y-with-floor\n\nhttps://www.kaggle.com/kokitanisaka/lstm-by-keras-with-unified-wi-fi-feats\n\nhttps://www.kaggle.com/oxzplvifi/indoor-gbm-postprocessing-xy-prediction\n\nhttps://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model\n\nhttps://www.kaggle.com/hiro5299834/wifi-features-with-lightgbm-kfold\n\nhttps://www.kaggle.com/ebinan92/time-series-rnn-xy-prediction\n\nWe released the following notebook to do \"Ensembling\". We used different methods in this notebook. For example, to eliminate bad answers in some stations, we used the \"decision rectangle\".\n\nhttps://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method\n\n****Cost Minimization****\n\nA magical notebook has been released that shows all the results can be improved. The address of this notebook is as follows:\n\nhttps://www.kaggle.com/saitodevel01/indoor-post-processing-by-cost-minimization\n\nYou can use this notebook before or after \"Ensembling\". Using this notebook, we also improved the score of each notebook before \"Ensembling\" and at the same time equalized the ['floor'] column for all of them. Our notebook address is as follows:\n\nhttps://www.kaggle.com/mehrankazeminia/1-3-indoor-navigation-cost-minimization\n\n****Push to hallway****\n\nSeveral notebooks have been published for this purpose. We used the excellent notebook below, which is called \"Snap to Grid\". We use this method only after \"Ensembling\". But if you use this method for each notebook before \"Ensembling\", several errors will accumulate and it is not clear that you can get a good result.\n\nhttps://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing\n\nThe author explains in this notebook that you can try numbers from 3 to 8 to determine the value for \"Threshold\". Of course, in our notebook, whose address is as follows, we used two values for \"Threshold\" at the same time to get a better result.\n\nhttps://www.kaggle.com/mehrankazeminia/3-3-g6-indoor-navigation-snap-to-grid\n\nIn the following notebook, a better way to determine the value of \"Threshold\" is done. The more numbers you try, the more accurate the number for the \"Threshold\" value. Of course, for different results, the value of \"Threshold\" also changes.\n\nhttps://www.kaggle.com/dragonzhang/3-3-g6-indoor-navigation-snap-to-grid\n\n****Fix the timestamps****\n\nIn the excellent notebook below, the problem is described in the [\"site_path_timestamp\"] column and suggestions are made:\n\nhttps://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask\n\nThen another notebook was released which is a development of the previous notebook and has excellent details.\n\nhttps://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage\n\nWe also used the above notebook to improve our results.\n\n****Good Luck****", "votes": null }, { "id": "1283294", "postDate": "04/24/2021 19:33:11", "content": "

Great. upvoted! @somayyehgholami 👍

", "rawMarkdown": "Great. upvoted! @somayyehgholami 👍", "votes": null }, { "id": "1284712", "postDate": "04/26/2021 08:12:16", "content": "

Thank you @yasserhessein

", "rawMarkdown": "Thank you @yasserhessein", "votes": null } ], "comments": [ { "id": 1283294, "author_name": "yasserhessein", "author_url": "", "post_date": "04/24/2021 19:33:11", "content": "

Great. upvoted! @somayyehgholami 👍

", "votes": null, "replies": [ { "id": 1284712, "author_name": "somayyehgholami", "author_url": "", "post_date": "04/26/2021 08:12:16", "content": "

Thank you @yasserhessein

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1260595": "I would like to mention a number of public notebooks that we have used. \nThanks to everyone who has published their notebooks publicly.\n\n****Fix the floor prediction****\n\nWe used the following great notebook for \"Floor prediction\". You can also try the ['floor'] column of the results of this notebook.\n\nhttps://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model\n\n****Ensembling****\n\nWe used a lot of public notebooks for \"Ensembling\" that scored very well. The addresses of some of them are as follows:\n\nhttps://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing\n\nhttps://www.kaggle.com/therocket290/lstm-unified-wi-fi-training-x-and-y-with-floor\n\nhttps://www.kaggle.com/kokitanisaka/lstm-by-keras-with-unified-wi-fi-feats\n\nhttps://www.kaggle.com/oxzplvifi/indoor-gbm-postprocessing-xy-prediction\n\nhttps://www.kaggle.com/nigelhenry/simple-99-accurate-floor-model\n\nhttps://www.kaggle.com/hiro5299834/wifi-features-with-lightgbm-kfold\n\nhttps://www.kaggle.com/ebinan92/time-series-rnn-xy-prediction\n\nWe released the following notebook to do \"Ensembling\". We used different methods in this notebook. For example, to eliminate bad answers in some stations, we used the \"decision rectangle\".\n\nhttps://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method\n\n****Cost Minimization****\n\nA magical notebook has been released that shows all the results can be improved. The address of this notebook is as follows:\n\nhttps://www.kaggle.com/saitodevel01/indoor-post-processing-by-cost-minimization\n\nYou can use this notebook before or after \"Ensembling\". Using this notebook, we also improved the score of each notebook before \"Ensembling\" and at the same time equalized the ['floor'] column for all of them. Our notebook address is as follows:\n\nhttps://www.kaggle.com/mehrankazeminia/1-3-indoor-navigation-cost-minimization\n\n****Push to hallway****\n\nSeveral notebooks have been published for this purpose. We used the excellent notebook below, which is called \"Snap to Grid\". We use this method only after \"Ensembling\". But if you use this method for each notebook before \"Ensembling\", several errors will accumulate and it is not clear that you can get a good result.\n\nhttps://www.kaggle.com/robikscube/indoor-navigation-snap-to-grid-post-processing\n\nThe author explains in this notebook that you can try numbers from 3 to 8 to determine the value for \"Threshold\". Of course, in our notebook, whose address is as follows, we used two values for \"Threshold\" at the same time to get a better result.\n\nhttps://www.kaggle.com/mehrankazeminia/3-3-g6-indoor-navigation-snap-to-grid\n\nIn the following notebook, a better way to determine the value of \"Threshold\" is done. The more numbers you try, the more accurate the number for the \"Threshold\" value. Of course, for different results, the value of \"Threshold\" also changes.\n\nhttps://www.kaggle.com/dragonzhang/3-3-g6-indoor-navigation-snap-to-grid\n\n****Fix the timestamps****\n\nIn the excellent notebook below, the problem is described in the [\"site_path_timestamp\"] column and suggestions are made:\n\nhttps://www.kaggle.com/jiweiliu/fix-the-timestamps-of-test-data-using-dask\n\nThen another notebook was released which is a development of the previous notebook and has excellent details.\n\nhttps://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage\n\nWe also used the above notebook to improve our results.\n\n****Good Luck****", "1283294": "Great. upvoted! @somayyehgholami 👍", "1284712": "Thank you @yasserhessein" }, "source": "meta" }