{ "id": 446125, "title": "Video available. Introductory webinar related to the competition", "url": "/competitions/child-mind-institute-detect-sleep-states/discussion/446125", "author_name": "", "post_date": "2023-10-10T12:19:03.680966600Z", "votes": 16, "comment_count": 5, "views": 0, "content": "

📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1
\n📖 Links to materials: see below

\n

Educational introductory webinar around the competition. Everybody welcome. Video record will be on youtube soon afterwards. Zoom link will be here ~5 minutes before the start.

\n

👨‍🔬 Oleg Khudyakov \"Introduction to Kaggle competition - Detect sleep states .\"
\n⌚️ Thursday 12 October, 18.00 (CET time)

\n

Add to Google Calendar

\n

Introduction and overview of the competition will be given. Parsing the raw data, including data labeling errors. Basic feature engineering. Application of gradient boosting and neural networks. Assumptions for postprocessing.

\n

Video records on data science and Kaggle: https://www.youtube.com/playlist?list=PL1GnA8S_asBANHU95kOPOPEh3moPm5uhw - subscribe !

\n

Zoom link will be here ~5 minutes before the start.

", "messages": [ { "id": "2476202", "postDate": "10/10/2023 12:19:03", "content": "

📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1
\n📖 Links to materials: see below

\n

Educational introductory webinar around the competition. Everybody welcome. Video record will be on youtube soon afterwards. Zoom link will be here ~5 minutes before the start.

\n

👨‍🔬 Oleg Khudyakov \"Introduction to Kaggle competition - Detect sleep states .\"
\n⌚️ Thursday 12 October, 18.00 (CET time)

\n

Add to Google Calendar

\n

Introduction and overview of the competition will be given. Parsing the raw data, including data labeling errors. Basic feature engineering. Application of gradient boosting and neural networks. Assumptions for postprocessing.

\n

Video records on data science and Kaggle: https://www.youtube.com/playlist?list=PL1GnA8S_asBANHU95kOPOPEh3moPm5uhw - subscribe !

\n

Zoom link will be here ~5 minutes before the start.

", "rawMarkdown": "📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1\n📖 Links to materials: [see below](https://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/446125#2479520) \n\nEducational introductory webinar around the competition. Everybody welcome. Video record will be on youtube soon afterwards. Zoom link will be here ~5 minutes before the start.\n\n👨‍🔬 Oleg Khudyakov \"Introduction to Kaggle competition - Detect sleep states .\"\n⌚️ Thursday 12 October, 18.00 (CET time)\n\n[Add to Google Calendar](https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20231012T160000Z%2F20231012T173000Z&details=Zoom%20link%20will%20be%20available%20in%20https%3A%2F%2Ft.me%2Fsberlogabig%20%20shortly%20before%20start%20of%20the%20talk.&text=%20%20Oleg%20Khudyakov%20%22Introduction%20to%20Kaggle%20competition%20-%20Detect%20sleep%20states%20%28https%3A%2F%2Fwww.kaggle.com%2Fcompetitions%2Fchild-mind-institute-detect-sleep-states%29.%22)\n\nIntroduction and overview of the competition will be given. Parsing the raw data, including data labeling errors. Basic feature engineering. Application of gradient boosting and neural networks. Assumptions for postprocessing.\n\n Video records on data science and Kaggle: https://www.youtube.com/playlist?list=PL1GnA8S_asBANHU95kOPOPEh3moPm5uhw - subscribe !\n\nZoom link will be here ~5 minutes before the start.", "votes": null }, { "id": "2479419", "postDate": "10/12/2023 15:57:10", "content": "

Zoom link:
\nhttps://us02web.zoom.us/j/85016786716?pwd=Q3hTYiswaWRlYmU4eCtBOFhlcHZTdz09

", "rawMarkdown": "Zoom link:\nhttps://us02web.zoom.us/j/85016786716?pwd=Q3hTYiswaWRlYmU4eCtBOFhlcHZTdz09", "votes": null }, { "id": "2479520", "postDate": "10/12/2023 17:01:06", "content": "

Useful notebooks discussed in the talk:
\n Data visualizationhttps://www.kaggle.com/code/patrick0302/viz-of-sleeping-time-series
\n Metric understandinghttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/43893
\n Quality of series data varieshttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/4447865.1. Nootebook with full series (no NaNs)https://www.kaggle.com/code/carlmcbrideellis/zzzs-make-small-starter-datasets-target5.2. Three times bigger dataset – series trimmed to first NaNhttps://www.kaggle.com/code/kaggledummie007/train-series-of-week-and-longer-w-o-nans

\n

Visualization of probabilities prediction and submissionhttps://www.kaggle.com/code/kaggledummie007/predictions-postprocessing-and-visualisation?scriptVersionId=146107561

\n

Great notebook with binary classification approach, fully commented, polars library used, submission file formed correctlyhttps://www.kaggle.com/code/lccburk/feature-engineering-and-random-forest-prediction

\n

The highest score public notebook at the moment - GRU RNN with score of 0,567 and «Critical point regression» approachhttps://www.kaggle.com/code/werus23/child-sleep-critical-point-regression

", "rawMarkdown": "Useful notebooks discussed in the talk:\n Data visualizationhttps://www.kaggle.com/code/patrick0302/viz-of-sleeping-time-series\n Metric understandinghttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/43893\n Quality of series data varieshttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/4447865.1. Nootebook with full series (no NaNs)https://www.kaggle.com/code/carlmcbrideellis/zzzs-make-small-starter-datasets-target5.2. Three times bigger dataset – series trimmed to first NaNhttps://www.kaggle.com/code/kaggledummie007/train-series-of-week-and-longer-w-o-nans\n\nVisualization of probabilities prediction and submissionhttps://www.kaggle.com/code/kaggledummie007/predictions-postprocessing-and-visualisation?scriptVersionId=146107561\n\nGreat notebook with binary classification approach, fully commented, polars library used, submission file formed correctlyhttps://www.kaggle.com/code/lccburk/feature-engineering-and-random-forest-prediction\n\n The highest score public notebook at the moment - GRU RNN with score of 0,567 and «Critical point regression» approachhttps://www.kaggle.com/code/werus23/child-sleep-critical-point-regression", "votes": null }, { "id": "2480734", "postDate": "10/13/2023 14:17:23", "content": "

📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1

", "rawMarkdown": "📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1", "votes": null }, { "id": "2487249", "postDate": "10/18/2023 13:24:30", "content": "

@kaggledummie007 Thanks for referring to my discussion post! The method I suggested produces the scores (probabilities) naturally. That's the beauty of it.

\n

I was referring to this paper by heatmap regression.

", "rawMarkdown": "kaggledummie007 Thanks for referring to my discussion post! The method I suggested produces the scores (probabilities) naturally. That's the beauty of it.\n\nI was referring to [this paper](https://openaccess.thecvf.com/content/CVPR2021/papers/Luo_Rethinking_the_Heatmap_Regression_for_Bottom-Up_Human_Pose_Estimation_CVPR_2021_paper.pdf) by heatmap regression.", "votes": null }, { "id": "2491382", "postDate": "10/21/2023 15:30:29", "content": "

thanks for sharing such a great idea!

", "rawMarkdown": "thanks for sharing such a great idea!", "votes": null } ], "comments": [ { "id": 2479419, "author_name": "alexandervc", "author_url": "", "post_date": "10/12/2023 15:57:10", "content": "

Zoom link:
\nhttps://us02web.zoom.us/j/85016786716?pwd=Q3hTYiswaWRlYmU4eCtBOFhlcHZTdz09

", "votes": null, "replies": [] }, { "id": 2479520, "author_name": "alexandervc", "author_url": "", "post_date": "10/12/2023 17:01:06", "content": "

Useful notebooks discussed in the talk:
\n Data visualizationhttps://www.kaggle.com/code/patrick0302/viz-of-sleeping-time-series
\n Metric understandinghttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/43893
\n Quality of series data varieshttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/4447865.1. Nootebook with full series (no NaNs)https://www.kaggle.com/code/carlmcbrideellis/zzzs-make-small-starter-datasets-target5.2. Three times bigger dataset – series trimmed to first NaNhttps://www.kaggle.com/code/kaggledummie007/train-series-of-week-and-longer-w-o-nans

\n

Visualization of probabilities prediction and submissionhttps://www.kaggle.com/code/kaggledummie007/predictions-postprocessing-and-visualisation?scriptVersionId=146107561

\n

Great notebook with binary classification approach, fully commented, polars library used, submission file formed correctlyhttps://www.kaggle.com/code/lccburk/feature-engineering-and-random-forest-prediction

\n

The highest score public notebook at the moment - GRU RNN with score of 0,567 and «Critical point regression» approachhttps://www.kaggle.com/code/werus23/child-sleep-critical-point-regression

", "votes": null, "replies": [] }, { "id": 2480734, "author_name": "alexandervc", "author_url": "", "post_date": "10/13/2023 14:17:23", "content": "

📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1

", "votes": null, "replies": [] }, { "id": 2487249, "author_name": "tolgadincer", "author_url": "", "post_date": "10/18/2023 13:24:30", "content": "

@kaggledummie007 Thanks for referring to my discussion post! The method I suggested produces the scores (probabilities) naturally. That's the beauty of it.

\n

I was referring to this paper by heatmap regression.

", "votes": null, "replies": [ { "id": 2491382, "author_name": "kaggledummie007", "author_url": "", "post_date": "10/21/2023 15:30:29", "content": "

thanks for sharing such a great idea!

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2476202": "📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1\n📖 Links to materials: [see below](https://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/446125#2479520) \n\nEducational introductory webinar around the competition. Everybody welcome. Video record will be on youtube soon afterwards. Zoom link will be here ~5 minutes before the start.\n\n👨‍🔬 Oleg Khudyakov \"Introduction to Kaggle competition - Detect sleep states .\"\n⌚️ Thursday 12 October, 18.00 (CET time)\n\n[Add to Google Calendar](https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20231012T160000Z%2F20231012T173000Z&details=Zoom%20link%20will%20be%20available%20in%20https%3A%2F%2Ft.me%2Fsberlogabig%20%20shortly%20before%20start%20of%20the%20talk.&text=%20%20Oleg%20Khudyakov%20%22Introduction%20to%20Kaggle%20competition%20-%20Detect%20sleep%20states%20%28https%3A%2F%2Fwww.kaggle.com%2Fcompetitions%2Fchild-mind-institute-detect-sleep-states%29.%22)\n\nIntroduction and overview of the competition will be given. Parsing the raw data, including data labeling errors. Basic feature engineering. Application of gradient boosting and neural networks. Assumptions for postprocessing.\n\n Video records on data science and Kaggle: https://www.youtube.com/playlist?list=PL1GnA8S_asBANHU95kOPOPEh3moPm5uhw - subscribe !\n\nZoom link will be here ~5 minutes before the start.", "2479419": "Zoom link:\nhttps://us02web.zoom.us/j/85016786716?pwd=Q3hTYiswaWRlYmU4eCtBOFhlcHZTdz09", "2479520": "Useful notebooks discussed in the talk:\n Data visualizationhttps://www.kaggle.com/code/patrick0302/viz-of-sleeping-time-series\n Metric understandinghttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/43893\n Quality of series data varieshttps://www.kaggle.com/competitions/child-mind-institute-detect-sleep-states/discussion/4447865.1. Nootebook with full series (no NaNs)https://www.kaggle.com/code/carlmcbrideellis/zzzs-make-small-starter-datasets-target5.2. Three times bigger dataset – series trimmed to first NaNhttps://www.kaggle.com/code/kaggledummie007/train-series-of-week-and-longer-w-o-nans\n\nVisualization of probabilities prediction and submissionhttps://www.kaggle.com/code/kaggledummie007/predictions-postprocessing-and-visualisation?scriptVersionId=146107561\n\nGreat notebook with binary classification approach, fully commented, polars library used, submission file formed correctlyhttps://www.kaggle.com/code/lccburk/feature-engineering-and-random-forest-prediction\n\n The highest score public notebook at the moment - GRU RNN with score of 0,567 and «Critical point regression» approachhttps://www.kaggle.com/code/werus23/child-sleep-critical-point-regression", "2480734": "📹 Video: https://youtu.be/RDnz9QASwaQ?si=KijFwYk94pAwYCk1", "2487249": "kaggledummie007 Thanks for referring to my discussion post! The method I suggested produces the scores (probabilities) naturally. That's the beauty of it.\n\nI was referring to [this paper](https://openaccess.thecvf.com/content/CVPR2021/papers/Luo_Rethinking_the_Heatmap_Regression_for_Bottom-Up_Human_Pose_Estimation_CVPR_2021_paper.pdf) by heatmap regression.", "2491382": "thanks for sharing such a great idea!" }, "source": "meta" }