{ "id": 221166, "title": "Indoor Comp: Where to get started and some useful discussions and EDA", "url": "/competitions/indoor-location-navigation/discussion/221166", "author_name": "", "post_date": "2021-02-21T16:46:30.367820200Z", "votes": 42, "comment_count": 4, "views": 0, "content": "

Hi All,

\n

[ updated with more discussion points on 1 April]

\n

My first competition on tis type of data-- and indoor positioning/navigation!

\n

I am trying to consolidate the key learnings from many discussions and codes. If you find anything interesting, please do comment and I will add it here. Hoping this helps us assimilate all the key learnings!

\n

1) LEAK and how to take advantage of it! Thanks @tomooinubushi for sharing with all!!
\nDiscussion: https://www.kaggle.com/c/indoor-location-navigation/discussion/228898
\nCode: https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage

\n

2)IMP discussions & code
\na) test set is very \"unbalanced\", some sites being over-represented
\nand site-specific model will do better
\nSource: https://www.kaggle.com/c/indoor-location-navigation/discussion/220835 @seuguh

\n

This theme is also discussed in https://www.kaggle.com/c/indoor-location-navigation/discussion/222391

\n

b) Also by @kokitanisaka how to create: one dataset for one model
\ndiscussion https://www.kaggle.com/c/indoor-location-navigation/discussion/222580
\nCode: https://www.kaggle.com/kokitanisaka/create-unified-wifi-features-example

\n

c) notebooks on using the locations to map and adjust the locations slightly .. (https://www.kaggle.com/c/indoor-location-navigation/discussion/221166)

\n

d) CV strategy @nooblife : https://www.kaggle.com/c/indoor-location-navigation/discussion/228199

\n

2) Correct floor mapping
\n(in evaluation page) :https://www.kaggle.com/c/indoor-location-navigation/overview/evaluation
\nand in discussion https://www.kaggle.com/c/indoor-location-navigation/discussion/220973 @jbomitchell

\n

3) Public vs Private dataset
\nNote that: This leaderboard is calculated with approximately 15% of the test data. The final results will be based on the other 85%, so the final standings may be different.

\n

As @nigelhenry highlights \"These are new paths in the test set, so you won't see it in the training set. You will however find the site in the training set.\" (https://www.kaggle.com/c/indoor-location-navigation/discussion/222142)
\nuseful for newbies to know…

\n

============
\n5) Dataset of all the 24 test(+train) sites in CSV format
\nThanks! @aatiffraz
\nhttps://www.kaggle.com/aatiffraz/indoor-location-navigation-tabular-data

\n

6) Papers related to Indoor Location & Navigation
\na) Great thread by @kingeee22
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/217874

\n

b) Awesome thread by @gpreda on resources
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/220546

\n

c) With privacy which we should also consider. Thanks @xhlulu !
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215979

\n

d) Another set by @saurabhshahane
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215223

\n

e) and more links by @deepakrajpurushothaman!! Thanks!
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215218

\n

f) Conferences & Competitions papers @satokiogiso
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215188

\n

7) Some EDA notebooks:
\na) Love all of @andradaolteanu 's notebooks,, they are so beautifully presented!
\nhttps://www.kaggle.com/andradaolteanu/indoor-navigation-complete-data-understanding

\n

b) Also, please look at EDA by @iamleonie
\nhttps://www.kaggle.com/iamleonie/intro-to-indoor-location-navigation

\n

c) and by @ihelon
\nhttps://www.kaggle.com/ihelon/indoor-location-exploratory-data-analysis

\n

I was amazed to see so many post papers and resources - was using it for my internal notes and felt may be useful to more of you!

", "messages": [ { "id": "1212837", "postDate": "02/21/2021 16:46:30", "content": "

Hi All,

\n

[ updated with more discussion points on 1 April]

\n

My first competition on tis type of data-- and indoor positioning/navigation!

\n

I am trying to consolidate the key learnings from many discussions and codes. If you find anything interesting, please do comment and I will add it here. Hoping this helps us assimilate all the key learnings!

\n

1) LEAK and how to take advantage of it! Thanks @tomooinubushi for sharing with all!!
\nDiscussion: https://www.kaggle.com/c/indoor-location-navigation/discussion/228898
\nCode: https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage

\n

2)IMP discussions & code
\na) test set is very \"unbalanced\", some sites being over-represented
\nand site-specific model will do better
\nSource: https://www.kaggle.com/c/indoor-location-navigation/discussion/220835 @seuguh

\n

This theme is also discussed in https://www.kaggle.com/c/indoor-location-navigation/discussion/222391

\n

b) Also by @kokitanisaka how to create: one dataset for one model
\ndiscussion https://www.kaggle.com/c/indoor-location-navigation/discussion/222580
\nCode: https://www.kaggle.com/kokitanisaka/create-unified-wifi-features-example

\n

c) notebooks on using the locations to map and adjust the locations slightly .. (https://www.kaggle.com/c/indoor-location-navigation/discussion/221166)

\n

d) CV strategy @nooblife : https://www.kaggle.com/c/indoor-location-navigation/discussion/228199

\n

2) Correct floor mapping
\n(in evaluation page) :https://www.kaggle.com/c/indoor-location-navigation/overview/evaluation
\nand in discussion https://www.kaggle.com/c/indoor-location-navigation/discussion/220973 @jbomitchell

\n

3) Public vs Private dataset
\nNote that: This leaderboard is calculated with approximately 15% of the test data. The final results will be based on the other 85%, so the final standings may be different.

\n

As @nigelhenry highlights \"These are new paths in the test set, so you won't see it in the training set. You will however find the site in the training set.\" (https://www.kaggle.com/c/indoor-location-navigation/discussion/222142)
\nuseful for newbies to know…

\n

============
\n5) Dataset of all the 24 test(+train) sites in CSV format
\nThanks! @aatiffraz
\nhttps://www.kaggle.com/aatiffraz/indoor-location-navigation-tabular-data

\n

6) Papers related to Indoor Location & Navigation
\na) Great thread by @kingeee22
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/217874

\n

b) Awesome thread by @gpreda on resources
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/220546

\n

c) With privacy which we should also consider. Thanks @xhlulu !
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215979

\n

d) Another set by @saurabhshahane
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215223

\n

e) and more links by @deepakrajpurushothaman!! Thanks!
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215218

\n

f) Conferences & Competitions papers @satokiogiso
\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215188

\n

7) Some EDA notebooks:
\na) Love all of @andradaolteanu 's notebooks,, they are so beautifully presented!
\nhttps://www.kaggle.com/andradaolteanu/indoor-navigation-complete-data-understanding

\n

b) Also, please look at EDA by @iamleonie
\nhttps://www.kaggle.com/iamleonie/intro-to-indoor-location-navigation

\n

c) and by @ihelon
\nhttps://www.kaggle.com/ihelon/indoor-location-exploratory-data-analysis

\n

I was amazed to see so many post papers and resources - was using it for my internal notes and felt may be useful to more of you!

", "rawMarkdown": "Hi All,\n\n[ updated with more discussion points on 1 April]\n\nMy first competition on tis type of data-- and indoor positioning/navigation!\n\nI am trying to consolidate the key learnings from many discussions and codes. If you find anything interesting, please do comment and I will add it here. Hoping this helps us assimilate all the key learnings! \n\n1) LEAK and how to take advantage of it! Thanks @tomooinubushi for sharing with all!!\nDiscussion: https://www.kaggle.com/c/indoor-location-navigation/discussion/228898\nCode: https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage\n\n2)IMP discussions & code\na) test set is very \"unbalanced\", some sites being over-represented \nand site-specific model will do better\nSource: https://www.kaggle.com/c/indoor-location-navigation/discussion/220835 @seuguh\n\nThis theme is also discussed in https://www.kaggle.com/c/indoor-location-navigation/discussion/222391 \n\nb) Also by @kokitanisaka how to create: one dataset for one model\ndiscussion https://www.kaggle.com/c/indoor-location-navigation/discussion/222580\nCode: https://www.kaggle.com/kokitanisaka/create-unified-wifi-features-example\n\nc) notebooks on using the locations to map and adjust the locations slightly .. (https://www.kaggle.com/c/indoor-location-navigation/discussion/221166) \n\nd) CV strategy @nooblife : https://www.kaggle.com/c/indoor-location-navigation/discussion/228199\n\n2) Correct floor mapping\n(in evaluation page) :https://www.kaggle.com/c/indoor-location-navigation/overview/evaluation\nand in discussion https://www.kaggle.com/c/indoor-location-navigation/discussion/220973 @jbomitchell\n\n3) Public vs Private dataset\nNote that: This leaderboard is calculated with approximately 15% of the test data. The final results will be based on the other 85%, so the final standings may be different.\n\nAs @nigelhenry highlights \"These are new paths in the test set, so you won't see it in the training set. You will however find the site in the training set.\" (https://www.kaggle.com/c/indoor-location-navigation/discussion/222142)\nuseful for newbies to know...\n\n============\n5) Dataset of all the 24 test(+train) sites in CSV format\nThanks! @aatiffraz\nhttps://www.kaggle.com/aatiffraz/indoor-location-navigation-tabular-data\n\n\n6) Papers related to Indoor Location & Navigation\na) Great thread by @kingeee22\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/217874\n\nb) Awesome thread by @gpreda on resources \nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/220546 \n\nc) With privacy which we should also consider. Thanks @xhlulu !\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215979\n\nd) Another set by @saurabhshahane\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215223\n\ne) and more links by @deepakrajpurushothaman!! Thanks!\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215218\n\nf) Conferences & Competitions papers @satokiogiso\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215188\n\n\n7) Some EDA notebooks:\na) Love all of @andradaolteanu 's notebooks,, they are so beautifully presented! \nhttps://www.kaggle.com/andradaolteanu/indoor-navigation-complete-data-understanding\n\nb) Also, please look at EDA by @iamleonie\nhttps://www.kaggle.com/iamleonie/intro-to-indoor-location-navigation\n\nc) and by @ihelon\nhttps://www.kaggle.com/ihelon/indoor-location-exploratory-data-analysis\n\n I was amazed to see so many post papers and resources - was using it for my internal notes and felt may be useful to more of you!", "votes": null }, { "id": "1214181", "postDate": "02/22/2021 16:27:27", "content": "

Thanks for compiling together, very useful for all of us!

", "rawMarkdown": "Thanks for compiling together, very useful for all of us!", "votes": null }, { "id": "1217036", "postDate": "02/24/2021 18:06:40", "content": "

thanks for sharing your insights and learning

", "rawMarkdown": "thanks for sharing your insights and learning", "votes": null }, { "id": "1229392", "postDate": "03/07/2021 10:05:28", "content": "

Thanks @kmldas for consolidation of useful links. It will be very helpful for people just starting in the competition

", "rawMarkdown": "Thanks @kmldas for consolidation of useful links. It will be very helpful for people just starting in the competition", "votes": null }, { "id": "1246875", "postDate": "03/21/2021 07:41:57", "content": "

thanks for sharing and really very useful.

", "rawMarkdown": "thanks for sharing and really very useful.", "votes": null } ], "comments": [ { "id": 1214181, "author_name": "saurabhbagchi", "author_url": "", "post_date": "02/22/2021 16:27:27", "content": "

Thanks for compiling together, very useful for all of us!

", "votes": null, "replies": [] }, { "id": 1217036, "author_name": "mohamadnawfal", "author_url": "", "post_date": "02/24/2021 18:06:40", "content": "

thanks for sharing your insights and learning

", "votes": null, "replies": [] }, { "id": 1229392, "author_name": "suryajrrafl", "author_url": "", "post_date": "03/07/2021 10:05:28", "content": "

Thanks @kmldas for consolidation of useful links. It will be very helpful for people just starting in the competition

", "votes": null, "replies": [] }, { "id": 1246875, "author_name": "somcool", "author_url": "", "post_date": "03/21/2021 07:41:57", "content": "

thanks for sharing and really very useful.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1212837": "Hi All,\n\n[ updated with more discussion points on 1 April]\n\nMy first competition on tis type of data-- and indoor positioning/navigation!\n\nI am trying to consolidate the key learnings from many discussions and codes. If you find anything interesting, please do comment and I will add it here. Hoping this helps us assimilate all the key learnings! \n\n1) LEAK and how to take advantage of it! Thanks @tomooinubushi for sharing with all!!\nDiscussion: https://www.kaggle.com/c/indoor-location-navigation/discussion/228898\nCode: https://www.kaggle.com/tomooinubushi/postprocessing-based-on-leakage\n\n2)IMP discussions & code\na) test set is very \"unbalanced\", some sites being over-represented \nand site-specific model will do better\nSource: https://www.kaggle.com/c/indoor-location-navigation/discussion/220835 @seuguh\n\nThis theme is also discussed in https://www.kaggle.com/c/indoor-location-navigation/discussion/222391 \n\nb) Also by @kokitanisaka how to create: one dataset for one model\ndiscussion https://www.kaggle.com/c/indoor-location-navigation/discussion/222580\nCode: https://www.kaggle.com/kokitanisaka/create-unified-wifi-features-example\n\nc) notebooks on using the locations to map and adjust the locations slightly .. (https://www.kaggle.com/c/indoor-location-navigation/discussion/221166) \n\nd) CV strategy @nooblife : https://www.kaggle.com/c/indoor-location-navigation/discussion/228199\n\n2) Correct floor mapping\n(in evaluation page) :https://www.kaggle.com/c/indoor-location-navigation/overview/evaluation\nand in discussion https://www.kaggle.com/c/indoor-location-navigation/discussion/220973 @jbomitchell\n\n3) Public vs Private dataset\nNote that: This leaderboard is calculated with approximately 15% of the test data. The final results will be based on the other 85%, so the final standings may be different.\n\nAs @nigelhenry highlights \"These are new paths in the test set, so you won't see it in the training set. You will however find the site in the training set.\" (https://www.kaggle.com/c/indoor-location-navigation/discussion/222142)\nuseful for newbies to know...\n\n============\n5) Dataset of all the 24 test(+train) sites in CSV format\nThanks! @aatiffraz\nhttps://www.kaggle.com/aatiffraz/indoor-location-navigation-tabular-data\n\n\n6) Papers related to Indoor Location & Navigation\na) Great thread by @kingeee22\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/217874\n\nb) Awesome thread by @gpreda on resources \nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/220546 \n\nc) With privacy which we should also consider. Thanks @xhlulu !\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215979\n\nd) Another set by @saurabhshahane\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215223\n\ne) and more links by @deepakrajpurushothaman!! Thanks!\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215218\n\nf) Conferences & Competitions papers @satokiogiso\nhttps://www.kaggle.com/c/indoor-location-navigation/discussion/215188\n\n\n7) Some EDA notebooks:\na) Love all of @andradaolteanu 's notebooks,, they are so beautifully presented! \nhttps://www.kaggle.com/andradaolteanu/indoor-navigation-complete-data-understanding\n\nb) Also, please look at EDA by @iamleonie\nhttps://www.kaggle.com/iamleonie/intro-to-indoor-location-navigation\n\nc) and by @ihelon\nhttps://www.kaggle.com/ihelon/indoor-location-exploratory-data-analysis\n\n I was amazed to see so many post papers and resources - was using it for my internal notes and felt may be useful to more of you!", "1214181": "Thanks for compiling together, very useful for all of us!", "1217036": "thanks for sharing your insights and learning", "1229392": "Thanks @kmldas for consolidation of useful links. It will be very helpful for people just starting in the competition", "1246875": "thanks for sharing and really very useful." }, "source": "meta" }