\nFirst, we have analyzed the best notebooks and tried to implement them. Some went well, some not, and from that, we tried to understand which are the best approaches to go forward; after a certain point, we have got our best scores in all of our submissions.
\nThen we have ensembled our best submissions and used this notebook. made by @ahmedewida.
Then we got a great notebook from @oxzplvifi , used that for the Floor prediction which given a better performance than our previous submission. Here is the notebook link. thank you very much sir for this notebook.
We have also tried the Comparative Method for our submissions but didn't work well for us though this was a great notebook by @mehrankazeminia. Here is the Notbook.
We have applied two postprocessing techniques and got the best. to make those interfaces we also have taken bits of help from the postprocessing techniques shared in code, we will try to make them public as soon as possible via a GitHub repository.
\nfor the postprocessing ideas, we really want to thank @iwatatakuya , @mehrankazeminia and @jwilliamhughdore.
Last but not the least we really want to thank @somayyehgholami for adding this wonderful discussion.
\nwe really want to thank and congratulate @vicensgaitan , @tomooinubushi , @chris62, @kuto0633 , @ymatioun , @higepon , @jiweiliu , @vicensgaitan and the top best kagglers, to help us to gain our knowledge, from your discussions, solutions and notebooks. Thank you very much, everyone.
",
"messages": [
{
"id": 1313864,
"postDate": "2021-05-18T18:47:45.180Z",
"content": "\nFirst, we have analyzed the best notebooks and tried to implement them. Some went well, some not, and from that, we tried to understand which are the best approaches to go forward; after a certain point, we have got our best scores in all of our submissions.
\nThen we have ensembled our best submissions and used this notebook. made by @ahmedewida.
Then we got a great notebook from @oxzplvifi , used that for the Floor prediction which given a better performance than our previous submission. Here is the notebook link. thank you very much sir for this notebook.
We have also tried the Comparative Method for our submissions but didn't work well for us though this was a great notebook by @mehrankazeminia. Here is the Notbook.
We have applied two postprocessing techniques and got the best. to make those interfaces we also have taken bits of help from the postprocessing techniques shared in code, we will try to make them public as soon as possible via a GitHub repository.
\nfor the postprocessing ideas, we really want to thank @iwatatakuya , @mehrankazeminia and @jwilliamhughdore.
Last but not the least we really want to thank @somayyehgholami for adding this wonderful discussion.
\nwe really want to thank and congratulate @vicensgaitan , @tomooinubushi , @chris62, @kuto0633 , @ymatioun , @higepon , @jiweiliu , @vicensgaitan and the top best kagglers, to help us to gain our knowledge, from your discussions, solutions and notebooks. Thank you very much, everyone.
",
"rawMarkdown": "#Hello everyone!! It was great to compete with the best kagglers, started our journey with this competition.\n\n######Our best submission is 4.84137 in private LB, we have chosen the wrong one, as we thought that will give better but it didn't 💔💔.\n Our work flow\n\n\n####Here is our small writeup about our approach to the best submission :\n\n- First, we have analyzed the best notebooks and tried to implement them. Some went well, some not, and from that, we tried to understand which are the best approaches to go forward; after a certain point, we have got our best scores in all of our submissions. \nThen we have ensembled our best submissions and used [this notebook](https://www.kaggle.com/ahmedewida/indoorlocation-ensembling). made by @ahmedewida.\n\n- Then we got a great notebook from @oxzplvifi , used that for the Floor prediction which given a better performance than our previous submission. Here is the [notebook link](https://www.kaggle.com/oxzplvifi/indoor-kmeans-gbm-floor-prediction/). thank you very much sir for this notebook.\n\n- We have also tried the Comparative Method for our submissions but didn't work well for us though this was a great notebook by @mehrankazeminia. Here is the [Notbook](https://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method).\n\n- We have applied two postprocessing techniques and got the best. to make those interfaces we also have taken bits of help from the postprocessing techniques shared in code, we will try to make them public as soon as possible via a GitHub repository.\nfor the postprocessing ideas, we really want to thank @iwatatakuya , @mehrankazeminia and @jwilliamhughdore.\n\n- Last but not the least we really want to thank @somayyehgholami for adding this wonderful [discussion](https://www.kaggle.com/c/indoor-location-navigation/discussion/230153).\nwe really want to thank and congratulate @vicensgaitan , @tomooinubushi , @chris62, @kuto0633 , @ymatioun , @higepon , @jiweiliu , @vicensgaitan and the top best kagglers, to help us to gain our knowledge, from your discussions, solutions and notebooks. Thank you very much, everyone. \n\n",
"votes": 15
}
],
"comments": [],
"raw_markdown_by_id": {
"1313864": "#Hello everyone!! It was great to compete with the best kagglers, started our journey with this competition.\n\n######Our best submission is 4.84137 in private LB, we have chosen the wrong one, as we thought that will give better but it didn't 💔💔.\n Our work flow\n\n\n####Here is our small writeup about our approach to the best submission :\n\n- First, we have analyzed the best notebooks and tried to implement them. Some went well, some not, and from that, we tried to understand which are the best approaches to go forward; after a certain point, we have got our best scores in all of our submissions. \nThen we have ensembled our best submissions and used [this notebook](https://www.kaggle.com/ahmedewida/indoorlocation-ensembling). made by @ahmedewida.\n\n- Then we got a great notebook from @oxzplvifi , used that for the Floor prediction which given a better performance than our previous submission. Here is the [notebook link](https://www.kaggle.com/oxzplvifi/indoor-kmeans-gbm-floor-prediction/). thank you very much sir for this notebook.\n\n- We have also tried the Comparative Method for our submissions but didn't work well for us though this was a great notebook by @mehrankazeminia. Here is the [Notbook](https://www.kaggle.com/mehrankazeminia/2-3-indoor-navigation-comparative-method).\n\n- We have applied two postprocessing techniques and got the best. to make those interfaces we also have taken bits of help from the postprocessing techniques shared in code, we will try to make them public as soon as possible via a GitHub repository.\nfor the postprocessing ideas, we really want to thank @iwatatakuya , @mehrankazeminia and @jwilliamhughdore.\n\n- Last but not the least we really want to thank @somayyehgholami for adding this wonderful [discussion](https://www.kaggle.com/c/indoor-location-navigation/discussion/230153).\nwe really want to thank and congratulate @vicensgaitan , @tomooinubushi , @chris62, @kuto0633 , @ymatioun , @higepon , @jiweiliu , @vicensgaitan and the top best kagglers, to help us to gain our knowledge, from your discussions, solutions and notebooks. Thank you very much, everyone. \n\n"
}
}![\"kkkk\"](\"https://user-images.githubusercontent.com/50532530/122804596-bae75a80-d2e5-11eb-9635-b19f7e5df157.PNG\")