I am a beginner in competitions. As a baseline, I am using the one provided by AMEDEO BIOLATTI. I've been trying to improve it with my own ideas, but I haven't seen any improvement on the leaderboard.
\nFeature Engineering:
\n・Addition of pressure, atmospheric density, and potential temperature at each layer.
Model Improvement:
\n・Addition of UNet and LSTM models.
Is my approach correct? I would appreciate any insights or tips for a beginner like me.
", "messages": [ { "id": 2909652, "postDate": "2024-07-07T07:29:28.937Z", "content": "I am a beginner in competitions. As a baseline, I am using the one provided by AMEDEO BIOLATTI. I've been trying to improve it with my own ideas, but I haven't seen any improvement on the leaderboard.
\nFeature Engineering:
\n・Addition of pressure, atmospheric density, and potential temperature at each layer.
Model Improvement:
\n・Addition of UNet and LSTM models.
Is my approach correct? I would appreciate any insights or tips for a beginner like me.
", "rawMarkdown": "I am a beginner in competitions. As a baseline, I am using the one provided by AMEDEO BIOLATTI. I've been trying to improve it with my own ideas, but I haven't seen any improvement on the leaderboard.\n\nFeature Engineering:\n・Addition of pressure, atmospheric density, and potential temperature at each layer.\n\nModel Improvement:\n・Addition of UNet and LSTM models.\n\nIs my approach correct? I would appreciate any insights or tips for a beginner like me.", "votes": 2 }, { "id": 2909662, "postDate": "2024-07-07T07:49:10.607Z", "content": "Here's a useful discussion on the topic, if not already checked: https://www.kaggle.com/competitions/leap-atmospheric-physics-ai-climsim/discussion/506984
", "rawMarkdown": "Here's a useful discussion on the topic, if not already checked: https://www.kaggle.com/competitions/leap-atmospheric-physics-ai-climsim/discussion/506984", "replies": [ { "id": 2909747, "postDate": "2024-07-07T09:11:43.620Z", "content": "Thank you. Should I focus more on improving the model rather than on feature engineering?
", "rawMarkdown": "Thank you. Should I focus more on improving the model rather than on feature engineering?", "replies": [ { "id": 2910462, "postDate": "2024-07-07T17:32:44.150Z", "content": "Both is good, as mentioned in the discussion, domain knowledge = feature engineering
", "rawMarkdown": "Both is good, as mentioned in the discussion, domain knowledge = feature engineering " } ] } ] } ], "comments": [ { "id": 2909662, "author_name": "g john rao", "author_url": "", "post_date": "2024-07-07T07:49:10.607000", "content": "Here's a useful discussion on the topic, if not already checked: https://www.kaggle.com/competitions/leap-atmospheric-physics-ai-climsim/discussion/506984
", "votes": 0, "replies": [ { "id": 2909747, "author_name": "yasukawatoyoki", "author_url": "", "post_date": "2024-07-07T09:11:43.620000", "content": "Thank you. Should I focus more on improving the model rather than on feature engineering?
", "votes": 0, "replies": [ { "id": 2910462, "author_name": "g john rao", "author_url": "", "post_date": "2024-07-07T17:32:44.150000", "content": "Both is good, as mentioned in the discussion, domain knowledge = feature engineering
", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2909652": "I am a beginner in competitions. As a baseline, I am using the one provided by AMEDEO BIOLATTI. I've been trying to improve it with my own ideas, but I haven't seen any improvement on the leaderboard.\n\nFeature Engineering:\n・Addition of pressure, atmospheric density, and potential temperature at each layer.\n\nModel Improvement:\n・Addition of UNet and LSTM models.\n\nIs my approach correct? I would appreciate any insights or tips for a beginner like me.", "2909662": "Here's a useful discussion on the topic, if not already checked: https://www.kaggle.com/competitions/leap-atmospheric-physics-ai-climsim/discussion/506984" } }