Hi! I'm experienced working with videos/images, but pretty bad at working with numerical data and feature engineering/selection. I'm curious to know what data has more meaning in this competition for you? What if there was only one type of data, which one would be more useful? And, if you are open to discuss, what are your CV scores based on only tracking vs only image data, and both? I have scores close to 0.7 for each type of data, and around 0.725 combined.
", "messages": [ { "id": 2107143, "postDate": "2023-01-19T15:47:10.930Z", "content": "Hi! I'm experienced working with videos/images, but pretty bad at working with numerical data and feature engineering/selection. I'm curious to know what data has more meaning in this competition for you? What if there was only one type of data, which one would be more useful? And, if you are open to discuss, what are your CV scores based on only tracking vs only image data, and both? I have scores close to 0.7 for each type of data, and around 0.725 combined.
", "rawMarkdown": "Hi! I'm experienced working with videos/images, but pretty bad at working with numerical data and feature engineering/selection. I'm curious to know what data has more meaning in this competition for you? What if there was only one type of data, which one would be more useful? And, if you are open to discuss, what are your CV scores based on only tracking vs only image data, and both? I have scores close to 0.7 for each type of data, and around 0.725 combined.", "votes": 17 }, { "id": 2152010, "postDate": "2023-02-20T14:22:25.677Z", "content": "cv0.768 lb0.771 ~~~
\nupdate:
\npb0.738πππ
impressive,I guess you use nn?
", "rawMarkdown": "impressive,I guess you use nn?", "votes": 1, "replies": [ { "id": 2152087, "postDate": "2023-02-20T14:58:54.287Z", "content": "Yes, NN > GBDT (in my conclusion).
", "rawMarkdown": "Yes, NN > GBDT (in my conclusion).", "votes": 6 } ] }, { "id": 2153224, "postDate": "2023-02-21T09:30:45.253Z", "content": "May I asking your XGB LB Score?
", "rawMarkdown": "May I asking your XGB LB Score?", "votes": 1, "replies": [ { "id": 2153251, "postDate": "2023-02-21T10:01:52.820Z", "content": "I have no time to testβ¦
", "rawMarkdown": "I have no time to test...", "votes": 1 } ] }, { "id": 2156557, "postDate": "2023-02-23T12:04:50.913Z", "content": "It's amazing ! How many layers of nn can get such good results ? We use 3 layers nn just get LB 0.68+.
", "rawMarkdown": "It's amazing ! How many layers of nn can get such good results ? We use 3 layers nn just get LB 0.68+.", "votes": 1, "replies": [ { "id": 2165558, "postDate": "2023-03-02T09:19:00.010Z", "content": "Just overfitting~
", "rawMarkdown": "Just overfitting~" } ] }, { "id": 2165552, "postDate": "2023-03-02T09:15:30.297Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 2154338, "postDate": "2023-02-22T00:23:26.447Z", "content": "CV: 0.740 / LB: 0.727 in 9 minutes
", "rawMarkdown": "CV: 0.740 / LB: 0.727 in 9 minutes", "votes": 4, "replies": [ { "id": 2154668, "postDate": "2023-02-22T07:19:02.420Z", "content": "Amazing. NN?
", "rawMarkdown": "Amazing. NN?", "votes": 1, "replies": [ { "id": 2155802, "postDate": "2023-02-22T23:06:00.870Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2155827, "postDate": "2023-02-22T23:51:42.313Z", "content": "No, it's single xgboost
", "rawMarkdown": "No, it's single xgboost" }, { "id": 2423024, "postDate": "2023-09-04T12:04:48.117Z", "content": "Wow thats good
", "rawMarkdown": "Wow thats good" } ] } ] }, { "id": 2151888, "postDate": "2023-02-20T12:49:12.153Z", "content": "Single lgb cv:0.732, lb:0.722 with only tabular data
", "rawMarkdown": "Single lgb cv:0.732, lb:0.722 with only tabular data", "votes": 1, "replies": [ { "id": 2151967, "postDate": "2023-02-20T13:44:57.150Z", "content": "do you think we can getting higher score by only tabular data?
", "rawMarkdown": "do you think we can getting higher score by only tabular data?", "votes": 1, "replies": [ { "id": 2152027, "postDate": "2023-02-20T14:37:23.853Z", "content": "definitely, but difficult to go to gold zone
", "rawMarkdown": "definitely, but difficult to go to gold zone", "votes": 1 } ] } ] }, { "id": 2148166, "postDate": "2023-02-17T07:37:25.310Z", "content": "Only got to 0.664 with tracking data only.
", "rawMarkdown": "Only got to 0.664 with tracking data only. ", "votes": 1 }, { "id": 2147906, "postDate": "2023-02-17T01:59:11.303Z", "content": "We get CV: 0.729 and LB:0.719 by only table data without image.
", "rawMarkdown": "We get CV: 0.729 and LB:0.719 by only table data without image.", "votes": 1 }, { "id": 2139663, "postDate": "2023-02-11T01:44:07.360Z", "content": "I use lgb and get cv:0.721 and lb:0.707 without image/videos. I'm still learning how to get more information from those \".mp4\" files.
", "rawMarkdown": "I use lgb and get cv:0.721 and lb:0.707 without image/videos. I'm still learning how to get more information from those \".mp4\" files.", "votes": 1 }, { "id": 2154223, "postDate": "2023-02-21T22:17:56.167Z", "content": "amazing, tabular looks well!
", "rawMarkdown": "amazing, tabular looks well!", "votes": -1 }, { "id": 2107888, "postDate": "2023-01-20T06:17:03.830Z", "content": "I am also interested in this question. By the way, I have no idea why there are feature data related to the videos, but there are still videos as our data. Is it means we can generate more features from the images?
", "rawMarkdown": "I am also interested in this question. By the way, I have no idea why there are feature data related to the videos, but there are still videos as our data. Is it means we can generate more features from the images?" } ], "comments": [ { "id": 2152010, "author_name": "hyd", "author_url": "", "post_date": "2023-02-20T14:22:25.677000", "content": "cv0.768 lb0.771 ~~~
\nupdate:
\npb0.738πππ
impressive,I guess you use nn?
", "votes": 1, "replies": [ { "id": 2152087, "author_name": "hyd", "author_url": "", "post_date": "2023-02-20T14:58:54.287000", "content": "Yes, NN > GBDT (in my conclusion).
", "votes": 6, "replies": [] } ] }, { "id": 2153224, "author_name": "Dewei Chen", "author_url": "", "post_date": "2023-02-21T09:30:45.253000", "content": "May I asking your XGB LB Score?
", "votes": 1, "replies": [ { "id": 2153251, "author_name": "hyd", "author_url": "", "post_date": "2023-02-21T10:01:52.820000", "content": "I have no time to testβ¦
", "votes": 1, "replies": [] } ] }, { "id": 2156557, "author_name": "NODE", "author_url": "", "post_date": "2023-02-23T12:04:50.913000", "content": "It's amazing ! How many layers of nn can get such good results ? We use 3 layers nn just get LB 0.68+.
", "votes": 1, "replies": [ { "id": 2165558, "author_name": "hyd", "author_url": "", "post_date": "2023-03-02T09:19:00.010000", "content": "Just overfitting~
", "votes": 0, "replies": [] } ] }, { "id": 2165552, "author_name": "", "author_url": "", "post_date": "2023-03-02T09:15:30.297000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 2154338, "author_name": "ONODERA", "author_url": "", "post_date": "2023-02-22T00:23:26.447000", "content": "CV: 0.740 / LB: 0.727 in 9 minutes
", "votes": 4, "replies": [ { "id": 2154668, "author_name": "nymfree", "author_url": "", "post_date": "2023-02-22T07:19:02.420000", "content": "Amazing. NN?
", "votes": 1, "replies": [ { "id": 2155802, "author_name": "", "author_url": "", "post_date": "2023-02-22T23:06:00.870000", "content": "", "votes": 0, "replies": [] }, { "id": 2155827, "author_name": "ONODERA", "author_url": "", "post_date": "2023-02-22T23:51:42.313000", "content": "No, it's single xgboost
", "votes": 0, "replies": [] }, { "id": 2423024, "author_name": "Bot_developer11", "author_url": "", "post_date": "2023-09-04T12:04:48.117000", "content": "Wow thats good
", "votes": 0, "replies": [] } ] } ] }, { "id": 2151888, "author_name": "senkin13", "author_url": "", "post_date": "2023-02-20T12:49:12.153000", "content": "Single lgb cv:0.732, lb:0.722 with only tabular data
", "votes": 1, "replies": [ { "id": 2151967, "author_name": "Dewei Chen", "author_url": "", "post_date": "2023-02-20T13:44:57.150000", "content": "do you think we can getting higher score by only tabular data?
", "votes": 1, "replies": [ { "id": 2152027, "author_name": "senkin13", "author_url": "", "post_date": "2023-02-20T14:37:23.853000", "content": "definitely, but difficult to go to gold zone
", "votes": 1, "replies": [] } ] } ] }, { "id": 2148166, "author_name": "nymfree", "author_url": "", "post_date": "2023-02-17T07:37:25.310000", "content": "Only got to 0.664 with tracking data only.
", "votes": 1, "replies": [] }, { "id": 2147906, "author_name": "Dewei Chen", "author_url": "", "post_date": "2023-02-17T01:59:11.303000", "content": "We get CV: 0.729 and LB:0.719 by only table data without image.
", "votes": 1, "replies": [] }, { "id": 2139663, "author_name": "Ethan", "author_url": "", "post_date": "2023-02-11T01:44:07.360000", "content": "I use lgb and get cv:0.721 and lb:0.707 without image/videos. I'm still learning how to get more information from those \".mp4\" files.
", "votes": 1, "replies": [] }, { "id": 2154223, "author_name": "", "author_url": "", "post_date": "2023-02-21T22:17:56.167000", "content": "amazing, tabular looks well!
", "votes": -1, "replies": [] }, { "id": 2107888, "author_name": "lifter", "author_url": "", "post_date": "2023-01-20T06:17:03.830000", "content": "I am also interested in this question. By the way, I have no idea why there are feature data related to the videos, but there are still videos as our data. Is it means we can generate more features from the images?
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2107143": "Hi! I'm experienced working with videos/images, but pretty bad at working with numerical data and feature engineering/selection. I'm curious to know what data has more meaning in this competition for you? What if there was only one type of data, which one would be more useful? And, if you are open to discuss, what are your CV scores based on only tracking vs only image data, and both? I have scores close to 0.7 for each type of data, and around 0.725 combined.", "2152010": "cv0.768 lb0.771 ~~~\nupdate:\npb0.738πππ", "2154338": "CV: 0.740 / LB: 0.727 in 9 minutes", "2151888": "Single lgb cv:0.732, lb:0.722 with only tabular data", "2148166": "Only got to 0.664 with tracking data only. ", "2147906": "We get CV: 0.729 and LB:0.719 by only table data without image.", "2139663": "I use lgb and get cv:0.721 and lb:0.707 without image/videos. I'm still learning how to get more information from those \".mp4\" files.", "2154223": "amazing, tabular looks well!", "2107888": "I am also interested in this question. By the way, I have no idea why there are feature data related to the videos, but there are still videos as our data. Is it means we can generate more features from the images?" } }