Thanks to the organizers and to everyone who share their ideas in public notebooks and discussion. I also needed to gain experience in analyzing single cell data and this competition help me gain a lot of knowledge and will become a precise experience.
\nI'm a postgraduate student and it's my first time to join the Kaggle competition, so if my organization in this nootbook was not clear and you want to know more other information, you can comment below or send me a private message. As a result of I didn't control the time very well, so the final plans were not been finished and submitted. My focus was on the feature engineering, some methods such as DCA, Magic, TruncatedSVD, FA, LDA and so on can make a positive effect in the final result. About the cross validation part, I divided the data by batch, but I don't think it's a good method for the private score. About the ensemble part, I selected the models including NN, Lightgbm, Catboost, Xgboost and Kernel ridge(For citeseq is ridge).
\nBecause the feature engineer's parts are not organized well, now I just to share the NN model structure which gained the best score among single models, though the final TruncaredSVD's parameters were changed a little.https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn.
", "messages": [ { "id": "2036198", "postDate": "11/19/2022 14:41:25", "content": "Thanks to the organizers and to everyone who share their ideas in public notebooks and discussion. I also needed to gain experience in analyzing single cell data and this competition help me gain a lot of knowledge and will become a precise experience.
\nI'm a postgraduate student and it's my first time to join the Kaggle competition, so if my organization in this nootbook was not clear and you want to know more other information, you can comment below or send me a private message. As a result of I didn't control the time very well, so the final plans were not been finished and submitted. My focus was on the feature engineering, some methods such as DCA, Magic, TruncatedSVD, FA, LDA and so on can make a positive effect in the final result. About the cross validation part, I divided the data by batch, but I don't think it's a good method for the private score. About the ensemble part, I selected the models including NN, Lightgbm, Catboost, Xgboost and Kernel ridge(For citeseq is ridge).
\nBecause the feature engineer's parts are not organized well, now I just to share the NN model structure which gained the best score among single models, though the final TruncaredSVD's parameters were changed a little.https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn.
", "rawMarkdown": "Thanks to the organizers and to everyone who share their ideas in public notebooks and discussion. I also needed to gain experience in analyzing single cell data and this competition help me gain a lot of knowledge and will become a precise experience.\n\nI'm a postgraduate student and it's my first time to join the Kaggle competition, so if my organization in this nootbook was not clear and you want to know more other information, you can comment below or send me a private message. As a result of I didn't control the time very well, so the final plans were not been finished and submitted. My focus was on the feature engineering, some methods such as DCA, Magic, TruncatedSVD, FA, LDA and so on can make a positive effect in the final result. About the cross validation part, I divided the data by batch, but I don't think it's a good method for the private score. About the ensemble part, I selected the models including NN, Lightgbm, Catboost, Xgboost and Kernel ridge(For citeseq is ridge).\n\nBecause the feature engineer's parts are not organized well, now I just to share the NN model structure which gained the best score among single models, though the final TruncaredSVD's parameters were changed a little.[https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn](https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn).", "votes": null }, { "id": "2036587", "postDate": "11/19/2022 23:36:40", "content": "Hi @songqizhou Thanks for sharing your solution. Does the Feature Engineering and other models (lgb, catboost etc) will be shared in future ?
", "rawMarkdown": "Hi @songqizhou Thanks for sharing your solution. Does the Feature Engineering and other models (lgb, catboost etc) will be shared in future ?", "votes": null }, { "id": "2036705", "postDate": "11/20/2022 05:07:16", "content": "Thanks for your attention, it's my honour to do that. I'll publish them in the github and post a link in this page when I organized them well.
", "rawMarkdown": "Thanks for your attention, it's my honour to do that. I'll publish them in the github and post a link in this page when I organized them well.", "votes": null }, { "id": "2041406", "postDate": "11/23/2022 23:03:52", "content": "Congratulations and thanks for posting your solution Barry.
", "rawMarkdown": "Congratulations and thanks for posting your solution Barry.", "votes": null }, { "id": "2041631", "postDate": "11/24/2022 05:11:24", "content": "Thank you ,MarĂlia Prata. đŸ˜„
", "rawMarkdown": "Thank you ,MarĂlia Prata. đŸ˜„", "votes": null }, { "id": "2084594", "postDate": "01/03/2023 15:58:35", "content": "Thanks for your sharing! Very helpful!
", "rawMarkdown": "Thanks for your sharing! Very helpful!", "votes": null } ], "comments": [ { "id": 2036587, "author_name": "bwhale", "author_url": "", "post_date": "11/19/2022 23:36:40", "content": "Hi @songqizhou Thanks for sharing your solution. Does the Feature Engineering and other models (lgb, catboost etc) will be shared in future ?
", "votes": null, "replies": [ { "id": 2036705, "author_name": "songqizhou", "author_url": "", "post_date": "11/20/2022 05:07:16", "content": "Thanks for your attention, it's my honour to do that. I'll publish them in the github and post a link in this page when I organized them well.
", "votes": null, "replies": [] } ] }, { "id": 2041406, "author_name": "mpwolke", "author_url": "", "post_date": "11/23/2022 23:03:52", "content": "Congratulations and thanks for posting your solution Barry.
", "votes": null, "replies": [ { "id": 2041631, "author_name": "songqizhou", "author_url": "", "post_date": "11/24/2022 05:11:24", "content": "Thank you ,MarĂlia Prata. đŸ˜„
", "votes": null, "replies": [] } ] }, { "id": 2084594, "author_name": "addoils", "author_url": "", "post_date": "01/03/2023 15:58:35", "content": "Thanks for your sharing! Very helpful!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2036198": "Thanks to the organizers and to everyone who share their ideas in public notebooks and discussion. I also needed to gain experience in analyzing single cell data and this competition help me gain a lot of knowledge and will become a precise experience.\n\nI'm a postgraduate student and it's my first time to join the Kaggle competition, so if my organization in this nootbook was not clear and you want to know more other information, you can comment below or send me a private message. As a result of I didn't control the time very well, so the final plans were not been finished and submitted. My focus was on the feature engineering, some methods such as DCA, Magic, TruncatedSVD, FA, LDA and so on can make a positive effect in the final result. About the cross validation part, I divided the data by batch, but I don't think it's a good method for the private score. About the ensemble part, I selected the models including NN, Lightgbm, Catboost, Xgboost and Kernel ridge(For citeseq is ridge).\n\nBecause the feature engineer's parts are not organized well, now I just to share the NN model structure which gained the best score among single models, though the final TruncaredSVD's parameters were changed a little.[https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn](https://www.kaggle.com/songqizhou/private-39th-public-4th-s-basical-single-model-nn).", "2036587": "Hi @songqizhou Thanks for sharing your solution. Does the Feature Engineering and other models (lgb, catboost etc) will be shared in future ?", "2036705": "Thanks for your attention, it's my honour to do that. I'll publish them in the github and post a link in this page when I organized them well.", "2041406": "Congratulations and thanks for posting your solution Barry.", "2041631": "Thank you ,MarĂlia Prata. đŸ˜„", "2084594": "Thanks for your sharing! Very helpful!" }, "source": "meta" }