{ "id": 82056, "title": "Can you share your best single model public/private lb?", "url": "/competitions/elo-merchant-category-recommendation/discussion/82056", "author_name": "Jianfei Wang", "post_date": "2019-02-27T02:50:28.611000", "votes": 8, "comment_count": 23, "views": 0, "content": "

Hi all:\n Our team best single model (such as just one lightgbm model )public lb is about 3.672 without any post process. As a new kaggler, i want to know more about this competition. Thank you!

", "messages": [ { "id": 479213, "postDate": "2019-02-27T02:50:28.613Z", "content": "

Hi all:\n Our team best single model (such as just one lightgbm model )public lb is about 3.672 without any post process. As a new kaggler, i want to know more about this competition. Thank you!

", "rawMarkdown": "Hi all:\n Our team best single model (such as just one lightgbm model )public lb is about 3.672 without any post process. As a new kaggler, i want to know more about this competition. Thank you!", "votes": 8 }, { "id": 479580, "postDate": "2019-02-27T08:57:23.837Z", "content": "

My best single model is a lightgbm with CV:3.6144,LB:3.668,PB:3.593

", "rawMarkdown": "My best single model is a lightgbm with CV:3.6144,LB:3.668,PB:3.593", "votes": 2, "replies": [ { "id": 479590, "postDate": "2019-02-27T09:06:57.877Z", "content": "

Very cool! thank you!

", "rawMarkdown": "Very cool! thank you!" } ] }, { "id": 479256, "postDate": "2019-02-27T03:15:26.043Z", "content": "

cv 3.629 public 3.680 private 3.603

", "rawMarkdown": "cv 3.629 public 3.680 private 3.603", "votes": 2, "replies": [ { "id": 479318, "postDate": "2019-02-27T04:14:24.877Z", "content": "

Good job!

", "rawMarkdown": "Good job!" }, { "id": 479321, "postDate": "2019-02-27T04:17:14.797Z", "content": "

one more question, your team single model public lb is 3.680? so how can you get 3.59x private lb. thank you!

", "rawMarkdown": "one more question, your team single model public lb is 3.680? so how can you get 3.59x private lb. thank you!" }, { "id": 479351, "postDate": "2019-02-27T04:54:18.830Z", "content": "

Only use stacking. our stack model score is cv 3.616,public 3.664,and private 3.593.

", "rawMarkdown": "Only use stacking. our stack model score is cv 3.616,public 3.664,and private 3.593." }, { "id": 479373, "postDate": "2019-02-27T05:13:49.367Z", "content": "

You are really good, with out 2 single model, ensemble get 3.666 public cv, but private only 3.599

", "rawMarkdown": "You are really good, with out 2 single model, ensemble get 3.666 public cv, but private only 3.599" } ] }, { "id": 479218, "postDate": "2019-02-27T02:57:45.540Z", "content": "

3.663 PLB 3.600 PriLb ( stacking )

", "rawMarkdown": "3.663 PLB 3.600 PriLb ( stacking )", "votes": 2, "replies": [ { "id": 479224, "postDate": "2019-02-27T03:00:22.647Z", "content": "

this is a stacking result, so how about your single model result?

", "rawMarkdown": "this is a stacking result, so how about your single model result?" }, { "id": 479239, "postDate": "2019-02-27T03:06:07.837Z", "content": "

3.667（LB, 372 features, cv3.641

", "rawMarkdown": "3.667（LB, 372 features, cv3.641", "votes": 1 }, { "id": 479248, "postDate": "2019-02-27T03:11:48.500Z", "content": "

很厉害！

", "rawMarkdown": "很厉害！" } ] }, { "id": 480143, "postDate": "2019-02-27T20:34:09.707Z", "content": "

Best single model (lgbm, 10-fold SKfold): 3.6434 CV, 3.682 public, 3.609 private

", "rawMarkdown": "Best single model (lgbm, 10-fold SKfold): 3.6434 CV, 3.682 public, 3.609 private" }, { "id": 479367, "postDate": "2019-02-27T05:07:22.273Z", "content": "

Unfortunately , I did not choose that submission but my best single model in private leaderboard is 3.605 in private, 3.692 in public. Very hard for me to figure out the performance of that single model in private leaderboard. Next time I think I should have a team.

", "rawMarkdown": "Unfortunately , I did not choose that submission but my best single model in private leaderboard is 3.605 in private, 3.692 in public. Very hard for me to figure out the performance of that single model in private leaderboard. Next time I think I should have a team. " }, { "id": 479296, "postDate": "2019-02-27T03:49:41.630Z", "content": "

I have used 399 features and get 3.689 in Public LB, 3.611 in Private LB and 3.645 in 5folds CV without stacking. \nAs a novice, I was surprised at the difference between public and private leadboard.

", "rawMarkdown": "I have used 399 features and get 3.689 in Public LB, 3.611 in Private LB and 3.645 in 5folds CV without stacking. \nAs a novice, I was surprised at the difference between public and private leadboard. ", "replies": [ { "id": 479319, "postDate": "2019-02-27T04:14:53.493Z", "content": "

me too.

", "rawMarkdown": "me too." } ] }, { "id": 479251, "postDate": "2019-02-27T03:13:14.950Z", "content": "

3.6352 3.674 3.604

", "rawMarkdown": " 3.6352 3.674 3.604", "replies": [ { "id": 479260, "postDate": "2019-02-27T03:18:07.107Z", "content": "

Thank you.

", "rawMarkdown": "Thank you. " } ] }, { "id": 479222, "postDate": "2019-02-27T02:59:54.473Z", "content": "

老哥互粉下嘛 upvote

", "rawMarkdown": "老哥互粉下嘛 upvote", "replies": [ { "id": 479243, "postDate": "2019-02-27T03:09:08.723Z", "content": "

哈哈，刚给你点了赞。

", "rawMarkdown": "哈哈，刚给你点了赞。" }, { "id": 479245, "postDate": "2019-02-27T03:09:51.263Z", "content": "

加个微信？ scuwangjianfei

", "rawMarkdown": "加个微信？ scuwangjianfei" }, { "id": 479262, "postDate": "2019-02-27T03:18:43.520Z", "content": "

我也想加大佬微信，2333333333

", "rawMarkdown": "我也想加大佬微信，2333333333" }, { "id": 479265, "postDate": "2019-02-27T03:23:18.860Z", "rawMarkdown": "", "isDeleted": true }, { "id": 479271, "postDate": "2019-02-27T03:28:37.707Z", "content": "

best single model 3.666 ！！！ 加了老哥

", "rawMarkdown": "best single model 3.666 ！！！ 加了老哥" } ] } ], "comments": [ { "id": 479580, "author_name": "senkin13", "author_url": "", "post_date": "2019-02-27T08:57:23.837000", "content": "

My best single model is a lightgbm with CV:3.6144,LB:3.668,PB:3.593

", "votes": 2, "replies": [ { "id": 479590, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T09:06:57.877000", "content": "

Very cool! thank you!

", "votes": 0, "replies": [] } ] }, { "id": 479256, "author_name": "rookzeno", "author_url": "", "post_date": "2019-02-27T03:15:26.043000", "content": "

cv 3.629 public 3.680 private 3.603

", "votes": 2, "replies": [ { "id": 479318, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T04:14:24.877000", "content": "

Good job!

", "votes": 0, "replies": [] }, { "id": 479321, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T04:17:14.797000", "content": "

one more question, your team single model public lb is 3.680? so how can you get 3.59x private lb. thank you!

", "votes": 0, "replies": [] }, { "id": 479351, "author_name": "rookzeno", "author_url": "", "post_date": "2019-02-27T04:54:18.830000", "content": "

Only use stacking. our stack model score is cv 3.616,public 3.664,and private 3.593.

", "votes": 0, "replies": [] }, { "id": 479373, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T05:13:49.367000", "content": "

You are really good, with out 2 single model, ensemble get 3.666 public cv, but private only 3.599

", "votes": 0, "replies": [] } ] }, { "id": 479218, "author_name": "Orange", "author_url": "", "post_date": "2019-02-27T02:57:45.540000", "content": "

3.663 PLB 3.600 PriLb ( stacking )

", "votes": 2, "replies": [ { "id": 479224, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T03:00:22.647000", "content": "

this is a stacking result, so how about your single model result?

", "votes": 0, "replies": [] }, { "id": 479239, "author_name": "process531", "author_url": "", "post_date": "2019-02-27T03:06:07.837000", "content": "

3.667（LB, 372 features, cv3.641

", "votes": 1, "replies": [] }, { "id": 479248, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T03:11:48.500000", "content": "

很厉害！

", "votes": 0, "replies": [] } ] }, { "id": 480143, "author_name": "Freddie Karlbom", "author_url": "", "post_date": "2019-02-27T20:34:09.707000", "content": "

Best single model (lgbm, 10-fold SKfold): 3.6434 CV, 3.682 public, 3.609 private

", "votes": 0, "replies": [] }, { "id": 479367, "author_name": "Icfstat", "author_url": "", "post_date": "2019-02-27T05:07:22.273000", "content": "

Unfortunately , I did not choose that submission but my best single model in private leaderboard is 3.605 in private, 3.692 in public. Very hard for me to figure out the performance of that single model in private leaderboard. Next time I think I should have a team.

", "votes": 0, "replies": [] }, { "id": 479296, "author_name": "heya5", "author_url": "", "post_date": "2019-02-27T03:49:41.630000", "content": "

I have used 399 features and get 3.689 in Public LB, 3.611 in Private LB and 3.645 in 5folds CV without stacking. \nAs a novice, I was surprised at the difference between public and private leadboard.

", "votes": 0, "replies": [ { "id": 479319, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T04:14:53.493000", "content": "

me too.

", "votes": 0, "replies": [] } ] }, { "id": 479251, "author_name": "Marcus Lin", "author_url": "", "post_date": "2019-02-27T03:13:14.950000", "content": "

3.6352 3.674 3.604

", "votes": 0, "replies": [ { "id": 479260, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T03:18:07.107000", "content": "

Thank you.

", "votes": 0, "replies": [] } ] }, { "id": 479222, "author_name": "Orange", "author_url": "", "post_date": "2019-02-27T02:59:54.473000", "content": "

老哥互粉下嘛 upvote

", "votes": 0, "replies": [ { "id": 479243, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T03:09:08.723000", "content": "

哈哈，刚给你点了赞。

", "votes": 0, "replies": [] }, { "id": 479245, "author_name": "Jianfei Wang", "author_url": "", "post_date": "2019-02-27T03:09:51.263000", "content": "

加个微信？ scuwangjianfei

", "votes": 0, "replies": [] }, { "id": 479262, "author_name": "process531", "author_url": "", "post_date": "2019-02-27T03:18:43.520000", "content": "

我也想加大佬微信，2333333333

", "votes": 0, "replies": [] }, { "id": 479265, "author_name": "", "author_url": "", "post_date": "2019-02-27T03:23:18.860000", "content": "", "votes": 0, "replies": [] }, { "id": 479271, "author_name": "Orange", "author_url": "", "post_date": "2019-02-27T03:28:37.707000", "content": "

best single model 3.666 ！！！ 加了老哥

", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "479213": "Hi all:\n Our team best single model (such as just one lightgbm model )public lb is about 3.672 without any post process. As a new kaggler, i want to know more about this competition. Thank you!", "479580": "My best single model is a lightgbm with CV:3.6144,LB:3.668,PB:3.593", "479256": "cv 3.629 public 3.680 private 3.603", "479218": "3.663 PLB 3.600 PriLb ( stacking )", "480143": "Best single model (lgbm, 10-fold SKfold): 3.6434 CV, 3.682 public, 3.609 private", "479367": "Unfortunately , I did not choose that submission but my best single model in private leaderboard is 3.605 in private, 3.692 in public. Very hard for me to figure out the performance of that single model in private leaderboard. Next time I think I should have a team. ", "479296": "I have used 399 features and get 3.689 in Public LB, 3.611 in Private LB and 3.645 in 5folds CV without stacking. \nAs a novice, I was surprised at the difference between public and private leadboard. ", "479251": " 3.6352 3.674 3.604", "479222": "老哥互粉下嘛 upvote" } }