I've understood the major part of protein sequencing and the terms (Not the file) used in the description, but I still do not understand what to do with certain files. (e.g. go-basic.obo, train_taxonomy.tsv, IA.txt, and maybe testsuperset_taxon_list.tsv). If anyone can help, please do.
", "messages": [ { "id": 2353665, "postDate": "2023-07-21T21:39:32.017Z", "content": "I've understood the major part of protein sequencing and the terms (Not the file) used in the description, but I still do not understand what to do with certain files. (e.g. go-basic.obo, train_taxonomy.tsv, IA.txt, and maybe testsuperset_taxon_list.tsv). If anyone can help, please do.
", "rawMarkdown": "I've understood the major part of protein sequencing and the terms (Not the file) used in the description, but I still do not understand what to do with certain files. (e.g. go-basic.obo, train_taxonomy.tsv, IA.txt, and maybe testsuperset_taxon_list.tsv). If anyone can help, please do.", "votes": 1 }, { "id": 2354603, "postDate": "2023-07-22T16:05:29.567Z", "content": "go-basic.obo provides the oriented graphical representation of the proteins system, links such as what protein is the parent/child of what other protein. It is technically used to produce the IA.txt that gives the weights used for the weighted mean precision score for the leaderboard : the deeper childs of the graph have a bigger weight than parents that are obviously more frequent in the dataset. So you can use it for optimization ends.
", "rawMarkdown": "go-basic.obo provides the oriented graphical representation of the proteins system, links such as what protein is the parent/child of what other protein. It is technically used to produce the IA.txt that gives the weights used for the weighted mean precision score for the leaderboard : the deeper childs of the graph have a bigger weight than parents that are obviously more frequent in the dataset. So you can use it for optimization ends. " }, { "id": 2354031, "postDate": "2023-07-22T07:41:21.800Z", "content": "You can check out various EDA and solution notebooks in the Code section of this competition which show different ways of solving this problem. If you're just getting started on Kaggle though, I personally think this competition is quite hard compared to others, so maybe not the best starting point.
", "rawMarkdown": "You can check out various EDA and solution notebooks in the Code section of this competition which show different ways of solving this problem. If you're just getting started on Kaggle though, I personally think this competition is quite hard compared to others, so maybe not the best starting point. ", "replies": [ { "id": 2354665, "postDate": "2023-07-22T17:21:02.300Z", "content": "What current competitions do you think I should participate in. And, yes indeed it is my first time in Kaggle.
", "rawMarkdown": "What current competitions do you think I should participate in. And, yes indeed it is my first time in Kaggle.", "replies": [ { "id": 2354733, "postDate": "2023-07-22T18:26:44.093Z", "content": "Depends on what you'd like to learn. ICR seems like a good starter comp, with a relatively simple problem formulation and a reasonable dataset size. On the deep learning side, both Commonlit and Kaggle LLM Science Exam should be good to start too. But if you're really interested in biology, then you might want to stick around to this one. It took me several weeks of study though to understand how to compete here. I think you'd need to read the papers shared by organizers, the public notebooks etc.
", "rawMarkdown": "Depends on what you'd like to learn. ICR seems like a good starter comp, with a relatively simple problem formulation and a reasonable dataset size. On the deep learning side, both Commonlit and Kaggle LLM Science Exam should be good to start too. But if you're really interested in biology, then you might want to stick around to this one. It took me several weeks of study though to understand how to compete here. I think you'd need to read the papers shared by organizers, the public notebooks etc. ", "votes": 2 }, { "id": 2357231, "postDate": "2023-07-24T15:54:08.867Z", "content": "I've understood the concept, and all of those things, I'm just new to tf.keras and scikit learn and any, like, python made-ready-for-you AI modules. Before now, I used to make simple AIs (usually regression models) using plain numpy. But here, that isn't that useful.
", "rawMarkdown": "I've understood the concept, and all of those things, I'm just new to tf.keras and scikit learn and any, like, python made-ready-for-you AI modules. Before now, I used to make simple AIs (usually regression models) using plain numpy. But here, that isn't that useful." } ] } ] }, { "id": 2359935, "postDate": "2023-07-26T13:02:35.043Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2354603, "author_name": "Henri Upton", "author_url": "", "post_date": "2023-07-22T16:05:29.567000", "content": "go-basic.obo provides the oriented graphical representation of the proteins system, links such as what protein is the parent/child of what other protein. It is technically used to produce the IA.txt that gives the weights used for the weighted mean precision score for the leaderboard : the deeper childs of the graph have a bigger weight than parents that are obviously more frequent in the dataset. So you can use it for optimization ends.
", "votes": 0, "replies": [] }, { "id": 2354031, "author_name": "Darek Kłeczek", "author_url": "", "post_date": "2023-07-22T07:41:21.800000", "content": "You can check out various EDA and solution notebooks in the Code section of this competition which show different ways of solving this problem. If you're just getting started on Kaggle though, I personally think this competition is quite hard compared to others, so maybe not the best starting point.
", "votes": 0, "replies": [ { "id": 2354665, "author_name": "Psychee", "author_url": "", "post_date": "2023-07-22T17:21:02.300000", "content": "What current competitions do you think I should participate in. And, yes indeed it is my first time in Kaggle.
", "votes": 0, "replies": [ { "id": 2354733, "author_name": "Darek Kłeczek", "author_url": "", "post_date": "2023-07-22T18:26:44.093000", "content": "Depends on what you'd like to learn. ICR seems like a good starter comp, with a relatively simple problem formulation and a reasonable dataset size. On the deep learning side, both Commonlit and Kaggle LLM Science Exam should be good to start too. But if you're really interested in biology, then you might want to stick around to this one. It took me several weeks of study though to understand how to compete here. I think you'd need to read the papers shared by organizers, the public notebooks etc.
", "votes": 2, "replies": [] }, { "id": 2357231, "author_name": "Psychee", "author_url": "", "post_date": "2023-07-24T15:54:08.867000", "content": "I've understood the concept, and all of those things, I'm just new to tf.keras and scikit learn and any, like, python made-ready-for-you AI modules. Before now, I used to make simple AIs (usually regression models) using plain numpy. But here, that isn't that useful.
", "votes": 0, "replies": [] } ] } ] }, { "id": 2359935, "author_name": "", "author_url": "", "post_date": "2023-07-26T13:02:35.043000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2353665": "I've understood the major part of protein sequencing and the terms (Not the file) used in the description, but I still do not understand what to do with certain files. (e.g. go-basic.obo, train_taxonomy.tsv, IA.txt, and maybe testsuperset_taxon_list.tsv). If anyone can help, please do.", "2354603": "go-basic.obo provides the oriented graphical representation of the proteins system, links such as what protein is the parent/child of what other protein. It is technically used to produce the IA.txt that gives the weights used for the weighted mean precision score for the leaderboard : the deeper childs of the graph have a bigger weight than parents that are obviously more frequent in the dataset. So you can use it for optimization ends. ", "2354031": "You can check out various EDA and solution notebooks in the Code section of this competition which show different ways of solving this problem. If you're just getting started on Kaggle though, I personally think this competition is quite hard compared to others, so maybe not the best starting point. ", "2359935": "" } }