Hello guys,\nI am new and am participating on this competition.\nLearning is the main goal here for me\nSo any one could help with resources on how to approach this sort of competition so that i can atleast get a direction to get started .\nI just need to understamd what should i work on to to atleast get a decent submission
", "messages": [ { "id": "952140", "postDate": "07/30/2020 16:40:02", "content": "Hello guys,\nI am new and am participating on this competition.\nLearning is the main goal here for me\nSo any one could help with resources on how to approach this sort of competition so that i can atleast get a direction to get started .\nI just need to understamd what should i work on to to atleast get a decent submission
", "rawMarkdown": "Hello guys,\nI am new and am participating on this competition.\nLearning is the main goal here for me\nSo any one could help with resources on how to approach this sort of competition so that i can atleast get a direction to get started .\nI just need to understamd what should i work on to to atleast get a decent submission", "votes": null }, { "id": "952200", "postDate": "07/30/2020 17:49:23", "content": "I'm a first-timer as well. One lesson learned so far is the following: start by submitting a very, very basic, simple model to see if your code works with the private data set (the data they score your model with), and add complexity little by little, testing each time to make sure your code submits. Just because the code runs without error on your end, doesn't mean it won't trip up on the private data set, and if it does you won't get any debugging info other than \"it didn't work\".
", "rawMarkdown": "I'm a first-timer as well. One lesson learned so far is the following: start by submitting a very, very basic, simple model to see if your code works with the private data set (the data they score your model with), and add complexity little by little, testing each time to make sure your code submits. Just because the code runs without error on your end, doesn't mean it won't trip up on the private data set, and if it does you won't get any debugging info other than \"it didn't work\".", "votes": null }, { "id": "952868", "postDate": "07/31/2020 09:23:02", "content": "Hi @nur988 \n- Read the competition Overview to get an idea about any particular competition\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview\n- Go through the Data section of the competition: It gives an idea about the provided dataset, columns, Expected submission format\nOnce you understand the competition Overview, Data, and Evaluation methods. You can start preparing for your solution.\n- Go through the Notebooks section to get an idea about how others are approaching the problem\n- Discussion Forum is very important. Keep track of the discussion section to get an idea about alternate methods, updates, suggestions.
\n\nAll the best!
", "rawMarkdown": "Hi @nur988 \n- Read the competition Overview to get an idea about any particular competition\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview\n- Go through the Data section of the competition: It gives an idea about the provided dataset, columns, Expected submission format\nOnce you understand the competition Overview, Data, and Evaluation methods. You can start preparing for your solution.\n- Go through the Notebooks section to get an idea about how others are approaching the problem\n- Discussion Forum is very important. Keep track of the discussion section to get an idea about alternate methods, updates, suggestions.\n\nAll the best!", "votes": null }, { "id": "953565", "postDate": "07/31/2020 21:46:47", "content": "I agree with @jagadish13 . I think going through notebooks offers a lot of insight. It will offer you more understanding of some concepts (in this case quantile regression, how to preprocess DICOM files etc.). @carlossouza made some great notebooks that I think could guide you in the right direction.\nGood luck!
", "rawMarkdown": "I agree with @jagadish13 . I think going through notebooks offers a lot of insight. It will offer you more understanding of some concepts (in this case quantile regression, how to preprocess DICOM files etc.). @carlossouza made some great notebooks that I think could guide you in the right direction.\nGood luck!", "votes": null }, { "id": "953951", "postDate": "08/01/2020 09:11:02", "content": "in addition to the valuable advice already given, start building (or borrowing one) your data pipeline as soon as possible. though you will inevitably start building your own like i am doing here for example. you want to make sure you can iterate fast and then gradually build it further upon your ideas
", "rawMarkdown": "in addition to the valuable advice already given, start building (or borrowing one) your data pipeline as soon as possible. though you will inevitably start building your own like i am doing [here](https://github.com/dron-dronych/OSIC-Pulmonary-Fibrosis-Progression) for example. you want to make sure you can **iterate fast** and then gradually build it further upon your ideas", "votes": null }, { "id": "953987", "postDate": "08/01/2020 09:40:00", "content": "👍
", "rawMarkdown": "👍", "votes": null } ], "comments": [ { "id": 952200, "author_name": "archimedus", "author_url": "", "post_date": "07/30/2020 17:49:23", "content": "I'm a first-timer as well. One lesson learned so far is the following: start by submitting a very, very basic, simple model to see if your code works with the private data set (the data they score your model with), and add complexity little by little, testing each time to make sure your code submits. Just because the code runs without error on your end, doesn't mean it won't trip up on the private data set, and if it does you won't get any debugging info other than \"it didn't work\".
", "votes": null, "replies": [] }, { "id": 952868, "author_name": "jagadish13", "author_url": "", "post_date": "07/31/2020 09:23:02", "content": "Hi @nur988 \n- Read the competition Overview to get an idea about any particular competition\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview\n- Go through the Data section of the competition: It gives an idea about the provided dataset, columns, Expected submission format\nOnce you understand the competition Overview, Data, and Evaluation methods. You can start preparing for your solution.\n- Go through the Notebooks section to get an idea about how others are approaching the problem\n- Discussion Forum is very important. Keep track of the discussion section to get an idea about alternate methods, updates, suggestions.
\n\nAll the best!
", "votes": null, "replies": [] }, { "id": 953565, "author_name": "sidbonshawn", "author_url": "", "post_date": "07/31/2020 21:46:47", "content": "I agree with @jagadish13 . I think going through notebooks offers a lot of insight. It will offer you more understanding of some concepts (in this case quantile regression, how to preprocess DICOM files etc.). @carlossouza made some great notebooks that I think could guide you in the right direction.\nGood luck!
", "votes": null, "replies": [ { "id": 953987, "author_name": "jagadish13", "author_url": "", "post_date": "08/01/2020 09:40:00", "content": "👍
", "votes": null, "replies": [] } ] }, { "id": 953951, "author_name": "dronych", "author_url": "", "post_date": "08/01/2020 09:11:02", "content": "in addition to the valuable advice already given, start building (or borrowing one) your data pipeline as soon as possible. though you will inevitably start building your own like i am doing here for example. you want to make sure you can iterate fast and then gradually build it further upon your ideas
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "952140": "Hello guys,\nI am new and am participating on this competition.\nLearning is the main goal here for me\nSo any one could help with resources on how to approach this sort of competition so that i can atleast get a direction to get started .\nI just need to understamd what should i work on to to atleast get a decent submission", "952200": "I'm a first-timer as well. One lesson learned so far is the following: start by submitting a very, very basic, simple model to see if your code works with the private data set (the data they score your model with), and add complexity little by little, testing each time to make sure your code submits. Just because the code runs without error on your end, doesn't mean it won't trip up on the private data set, and if it does you won't get any debugging info other than \"it didn't work\".", "952868": "Hi @nur988 \n- Read the competition Overview to get an idea about any particular competition\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview\n- Go through the Data section of the competition: It gives an idea about the provided dataset, columns, Expected submission format\nOnce you understand the competition Overview, Data, and Evaluation methods. You can start preparing for your solution.\n- Go through the Notebooks section to get an idea about how others are approaching the problem\n- Discussion Forum is very important. Keep track of the discussion section to get an idea about alternate methods, updates, suggestions.\n\nAll the best!", "953565": "I agree with @jagadish13 . I think going through notebooks offers a lot of insight. It will offer you more understanding of some concepts (in this case quantile regression, how to preprocess DICOM files etc.). @carlossouza made some great notebooks that I think could guide you in the right direction.\nGood luck!", "953951": "in addition to the valuable advice already given, start building (or borrowing one) your data pipeline as soon as possible. though you will inevitably start building your own like i am doing [here](https://github.com/dron-dronych/OSIC-Pulmonary-Fibrosis-Progression) for example. you want to make sure you can **iterate fast** and then gradually build it further upon your ideas", "953987": "👍" }, "source": "meta" }