Thank you to the organizers for organizing this competition, and thank you to all the participants for their efforts. I have learned a lot from this competition.
\nHere is my solution, with a score of 2.4067 on the public leaderboard.Below, I will briefly explain my solution.
\nAlthough the competition requires us to predict the next hour's data based on 6 hours of data, analyzing the correlation between the target and time reveals that the only time that truly correlates with the target is the 20 minutes closest to it. Data that is distant in time may become noise. So here we have transformed the problem into 'predicting the next hour's data based on the data from the previous x minutes'. After the final experiment, it was found that only about 45 minutes are enough.
\nI have also constructed the test data here, and I am not sure if the competition allows it, but I can guarantee that performing this operation on the training data will definitely improve the performance of the model. On the one hand, it removes noise, and on the other hand, it increases the amount of data.
\n2.Feature Engineer
\nA lot of feature engineering has been done here, such as the common sine cosine transform for temporal features, constructing cross features (addition, subtraction, multiplication, and division) in the original features, and calculating the temporal diff features.
\n3.ensemble
\nIn addition to using three models myself during model training, I also referenced open-source code from others.Here is the code.
\n\nsubmission=pd.read_csv()\n\nsubmission2=pd.read_csv()\n\nsubmission3=pd.read_csv()\nsubmission[]=*submission[]+*submission2[]+*submission3[]\nsubmission.to_csv(,index=)\nsubmission.head()\n4.Yunbase
\nWelcome to use Yunbase in other competitions.
Thank you to the organizers for organizing this competition, and thank you to all the participants for their efforts. I have learned a lot from this competition.
\nHere is my solution, with a score of 2.4067 on the public leaderboard.Below, I will briefly explain my solution.
\nAlthough the competition requires us to predict the next hour's data based on 6 hours of data, analyzing the correlation between the target and time reveals that the only time that truly correlates with the target is the 20 minutes closest to it. Data that is distant in time may become noise. So here we have transformed the problem into 'predicting the next hour's data based on the data from the previous x minutes'. After the final experiment, it was found that only about 45 minutes are enough.
\nI have also constructed the test data here, and I am not sure if the competition allows it, but I can guarantee that performing this operation on the training data will definitely improve the performance of the model. On the one hand, it removes noise, and on the other hand, it increases the amount of data.
\n2.Feature Engineer
\nA lot of feature engineering has been done here, such as the common sine cosine transform for temporal features, constructing cross features (addition, subtraction, multiplication, and division) in the original features, and calculating the temporal diff features.
\n3.ensemble
\nIn addition to using three models myself during model training, I also referenced open-source code from others.Here is the code.
\n\nsubmission=pd.read_csv()\n\nsubmission2=pd.read_csv()\n\nsubmission3=pd.read_csv()\nsubmission[]=*submission[]+*submission2[]+*submission3[]\nsubmission.to_csv(,index=)\nsubmission.head()\n4.Yunbase
\nWelcome to use Yunbase in other competitions.
Congratulation @yunsuxiaozi ! Happy to see you in top 5.
", "rawMarkdown": "Congratulation @yunsuxiaozi ! Happy to see you in top 5. ", "votes": 1 }, { "id": 3070145, "postDate": "2024-12-12T10:23:42.193Z", "content": "Hi @yunsuxiaozi,
\nI've contacted you via email (using the one you attached to the submission) about organising the competition prize as part of your winning submission. Please let me know if you have received this email and if you are interested in claiming your prize.
", "rawMarkdown": "Hi @yunsuxiaozi,\n\nI've contacted you via email (using the one you attached to the submission) about organising the competition prize as part of your winning submission. Please let me know if you have received this email and if you are interested in claiming your prize.", "replies": [ { "id": 3070287, "postDate": "2024-12-12T14:01:28.800Z", "content": "I've noticed that the prize for this competition is a gift card. Is it only usable for specific purposes? In that case, it's not very useful to me, so I don't need this prize.
", "rawMarkdown": "I've noticed that the prize for this competition is a gift card. Is it only usable for specific purposes? In that case, it's not very useful to me, so I don't need this prize.", "replies": [ { "id": 3070321, "postDate": "2024-12-12T14:46:29.470Z", "content": "My understanding is that the global reward codes can be converted into Virtual Visa cards in local currencies for most countries and then used anywhere that accepts Visa.
", "rawMarkdown": "My understanding is that the global reward codes can be converted into Virtual Visa cards in local currencies for most countries and then used anywhere that accepts Visa." }, { "id": 3070847, "postDate": "2024-12-13T05:21:12.977Z", "rawMarkdown": "", "isDeleted": true } ] } ] } ], "comments": [ { "id": 3060667, "author_name": "Sheikh Muhammad Abdullah", "author_url": "", "post_date": "2024-12-02T00:40:14.727000", "content": "Congratulation @yunsuxiaozi ! Happy to see you in top 5.
", "votes": 1, "replies": [] }, { "id": 3070145, "author_name": "Sam Gordon James", "author_url": "", "post_date": "2024-12-12T10:23:42.193000", "content": "Hi @yunsuxiaozi,
\nI've contacted you via email (using the one you attached to the submission) about organising the competition prize as part of your winning submission. Please let me know if you have received this email and if you are interested in claiming your prize.
", "votes": 0, "replies": [ { "id": 3070287, "author_name": "yunsuxiaozi", "author_url": "", "post_date": "2024-12-12T14:01:28.800000", "content": "I've noticed that the prize for this competition is a gift card. Is it only usable for specific purposes? In that case, it's not very useful to me, so I don't need this prize.
", "votes": 0, "replies": [ { "id": 3070321, "author_name": "Sam Gordon James", "author_url": "", "post_date": "2024-12-12T14:46:29.470000", "content": "My understanding is that the global reward codes can be converted into Virtual Visa cards in local currencies for most countries and then used anywhere that accepts Visa.
", "votes": 0, "replies": [] }, { "id": 3070847, "author_name": "", "author_url": "", "post_date": "2024-12-13T05:21:12.977000", "content": "", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "3059690": "Thank you to the organizers for organizing this competition, and thank you to all the participants for their efforts. I have learned a lot from this competition.\n\nHere is my solution, with a score of 2.4067 on the public leaderboard.Below, I will briefly explain my solution.\n\n1. Problem conversion (obtaining more data)\n\nAlthough the competition requires us to predict the next hour's data based on 6 hours of data, analyzing the correlation between the target and time reveals that the only time that truly correlates with the target is the 20 minutes closest to it. Data that is distant in time may become noise. So here we have transformed the problem into 'predicting the next hour's data based on the data from the previous x minutes'. After the final experiment, it was found that only about 45 minutes are enough.\n\nI have also constructed the test data here, and I am not sure if the competition allows it, but I can guarantee that performing this operation on the training data will definitely improve the performance of the model. On the one hand, it removes noise, and on the other hand, it increases the amount of data.\n\n2.Feature Engineer\n\nA lot of feature engineering has been done here, such as the common sine cosine transform for temporal features, constructing cross features (addition, subtraction, multiplication, and division) in the original features, and calculating the temporal diff features.\n\n \n3.ensemble\n\nIn addition to using three models myself during model training, I also referenced open-source code from others.Here is the code.\n\n```python\n#v1\nsubmission=pd.read_csv(\"/kaggle/input/brist1d-1126-yunbase/yunbase.csv\")\n#v3\nsubmission2=pd.read_csv(\"/kaggle/input/brist1d-stack-regressor-tabnet/submission.csv\")\n#ensemble v1\nsubmission3=pd.read_csv(\"/kaggle/input/brist1d-ensemble-3-step-towards-hyperspace/submission.csv\")\nsubmission['bg+1:00']=0.7*submission['bg+1:00']+0.1*submission2['bg+1:00']+0.2*submission3['bg+1:00']\nsubmission.to_csv(\"1128ensemble.csv\",index=None)\nsubmission.head()\n```\n\n4.Yunbase\nWelcome to use Yunbase in other competitions.", "3060667": "Congratulation @yunsuxiaozi ! Happy to see you in top 5. ", "3070145": "Hi @yunsuxiaozi,\n\nI've contacted you via email (using the one you attached to the submission) about organising the competition prize as part of your winning submission. Please let me know if you have received this email and if you are interested in claiming your prize." } }