{ "id": 82034, "title": "My First Solo Gold ", "url": "/competitions/elo-merchant-category-recommendation/discussion/82034", "author_name": "Anthony Chiu", "post_date": "2019-02-27T00:06:33.370000", "votes": 61, "comment_count": 81, "views": 0, "content": "

I was expecting a bronze and wishing for a silver. I was shocked in the morning.\nThanks all discussions and kernels here. It was a frustrating game. I have learned a lot here.
\nDetails of what I did: \nhttps://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82166

\n\n

I think CV wins in this game right?

\n\n

\"enter

", "messages": [ { "id": 479034, "postDate": "2019-02-27T00:06:33.370Z", "content": "

I was expecting a bronze and wishing for a silver. I was shocked in the morning.\nThanks all discussions and kernels here. It was a frustrating game. I have learned a lot here.
\nDetails of what I did: \nhttps://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82166

\n\n

I think CV wins in this game right?

\n\n

\"enter

", "rawMarkdown": "I was expecting a bronze and wishing for a silver. I was shocked in the morning.\nThanks all discussions and kernels here. It was a frustrating game. I have learned a lot here. \nDetails of what I did: \nhttps://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82166\n\nI think CV wins in this game right?\n\n![enter image description here][1]\n\n\n [1]: https://i.imgur.com/PM7UxW3.png\n", "votes": 61 }, { "id": 480493, "postDate": "2019-02-28T08:40:09.360Z", "content": "

Congratulations!

", "rawMarkdown": "Congratulations!", "votes": 1 }, { "id": 479995, "postDate": "2019-02-27T16:56:26.120Z", "content": "

Congratulations.

", "rawMarkdown": "Congratulations.", "votes": 1 }, { "id": 479880, "postDate": "2019-02-27T14:29:56.410Z", "content": "

congratulations (y)

", "rawMarkdown": "congratulations (y)", "votes": 1 }, { "id": 479795, "postDate": "2019-02-27T12:35:27.553Z", "content": "

Congratulations for your first gold medal

", "rawMarkdown": "Congratulations for your first gold medal", "votes": 1 }, { "id": 479758, "postDate": "2019-02-27T11:58:48.243Z", "content": "

Congratulations.

", "rawMarkdown": "Congratulations.", "votes": 1 }, { "id": 479737, "postDate": "2019-02-27T11:31:10.763Z", "content": "

Congrats

", "rawMarkdown": "Congrats", "votes": 1 }, { "id": 479734, "postDate": "2019-02-27T11:29:00.230Z", "content": "

Congrats!!

", "rawMarkdown": "Congrats!!", "votes": 1 }, { "id": 479636, "postDate": "2019-02-27T09:56:56.910Z", "content": "

Congratulations on your first win

", "rawMarkdown": "Congratulations on your first win ", "votes": 1 }, { "id": 479576, "postDate": "2019-02-27T08:54:31.010Z", "content": "

congratulations !!! well deserved.

", "rawMarkdown": "congratulations !!! well deserved.", "votes": 1 }, { "id": 479573, "postDate": "2019-02-27T08:54:11.820Z", "content": "

Congratulations

", "rawMarkdown": "Congratulations", "votes": 1 }, { "id": 479567, "postDate": "2019-02-27T08:49:58.750Z", "content": "

Congratulation ! I'm looking forward to see your work.

", "rawMarkdown": "Congratulation ! I'm looking forward to see your work. ", "votes": 1 }, { "id": 479478, "postDate": "2019-02-27T07:18:05.200Z", "content": "

Congratulations @kingychiu Keep up the great work going.

", "rawMarkdown": "Congratulations @kingychiu Keep up the great work going.", "votes": 1 }, { "id": 479450, "postDate": "2019-02-27T06:36:24.653Z", "content": "

Congratulations!!! and could you please share more about your solution, such as feature engineering, models, etc..

", "rawMarkdown": "Congratulations!!! and could you please share more about your solution, such as feature engineering, models, etc..", "votes": 1 }, { "id": 479404, "postDate": "2019-02-27T05:37:14.523Z", "content": "

Congratulations!!! . Keep up your good work and you will get the grandmaster tier soon! . I was dreaming of getting a solo gold medal, I took high risk with post processing and I did not select my best single models as finals submissions which would have been at least very near to gold positions. This situation tells me that I still have to improve my skills and work harder to get that fantastic solo gold medal ! . Congratulations again !

", "rawMarkdown": "Congratulations!!! . Keep up your good work and you will get the grandmaster tier soon! . I was dreaming of getting a solo gold medal, I took high risk with post processing and I did not select my best single models as finals submissions which would have been at least very near to gold positions. This situation tells me that I still have to improve my skills and work harder to get that fantastic solo gold medal ! . Congratulations again !", "votes": 1, "replies": [ { "id": 479993, "postDate": "2019-02-27T16:53:57.993Z", "content": "

Thanks for your kind words. I was so scared about manual post-processing and haven't done it. We are all learning here :)

", "rawMarkdown": "Thanks for your kind words. I was so scared about manual post-processing and haven't done it. We are all learning here :)", "votes": 1 } ] }, { "id": 479369, "postDate": "2019-02-27T05:10:44.867Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!\n\n", "votes": 1 }, { "id": 479355, "postDate": "2019-02-27T04:58:15.560Z", "content": "

Congratulations, Guess you have held on to you CV, Could you share how was CV selecting done by you like, number of rows in Local CV, Random rows or first/last n rows etc..

", "rawMarkdown": "Congratulations, Guess you have held on to you CV, Could you share how was CV selecting done by you like, number of rows in Local CV, Random rows or first/last n rows etc..", "votes": 1, "replies": [ { "id": 479487, "postDate": "2019-02-27T07:30:09.200Z", "content": "

Thanks, what do you mean by CV selecting? I simply picked the highest CV and the highest LB as the final 2 submissions.

", "rawMarkdown": "Thanks, what do you mean by CV selecting? I simply picked the highest CV and the highest LB as the final 2 submissions.", "votes": 1 } ] }, { "id": 479337, "postDate": "2019-02-27T04:35:27.190Z", "content": "

Congrats...

", "rawMarkdown": "Congrats...", "votes": 1 }, { "id": 479317, "postDate": "2019-02-27T04:14:14.140Z", "content": "

Congratulations ! @nlgn.\nWe hope you would share your strategy and insights with the community. Would love to learn from your work.

", "rawMarkdown": "Congratulations ! @nlgn.\nWe hope you would share your strategy and insights with the community. Would love to learn from your work. ", "votes": 1 }, { "id": 479312, "postDate": "2019-02-27T04:08:02.950Z", "content": "

Congrats @nlgn. It was indeed a tricky competition. And I'm happy that I trusted my CV. CV FTW!

", "rawMarkdown": "Congrats @nlgn. It was indeed a tricky competition. And I'm happy that I trusted my CV. CV FTW!", "votes": 1 }, { "id": 479292, "postDate": "2019-02-27T03:49:00.007Z", "content": "

Congrats!

", "rawMarkdown": "Congrats! ", "votes": 1 }, { "id": 479284, "postDate": "2019-02-27T03:38:16.573Z", "content": "

Congrats! Trusting CV wins

", "rawMarkdown": "Congrats! Trusting CV wins", "votes": 1 }, { "id": 479233, "postDate": "2019-02-27T03:05:01.553Z", "content": "

say nothing more, congrats!👍

", "rawMarkdown": "say nothing more, congrats!👍", "votes": 1 }, { "id": 479175, "postDate": "2019-02-27T02:10:50.977Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!", "votes": 1 }, { "id": 479158, "postDate": "2019-02-27T01:51:02.433Z", "content": "

very amazing, your blending is quite stable

", "rawMarkdown": "very amazing, your blending is quite stable", "votes": 1 }, { "id": 479143, "postDate": "2019-02-27T01:40:25.293Z", "content": "

Congrats! Trust your CV is first rule, LOL.

", "rawMarkdown": "Congrats! Trust your CV is first rule, LOL.", "votes": 1 }, { "id": 479138, "postDate": "2019-02-27T01:36:19.910Z", "content": "

Congratulations. Would love to see an writeup of your solution.

", "rawMarkdown": "Congratulations. Would love to see an writeup of your solution.", "votes": 1 }, { "id": 479136, "postDate": "2019-02-27T01:34:32.900Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!", "votes": 1 }, { "id": 479133, "postDate": "2019-02-27T01:31:41.287Z", "content": "

more is coming

", "rawMarkdown": "more is coming", "votes": 1 }, { "id": 479120, "postDate": "2019-02-27T01:19:38.657Z", "content": "

Congrats to you !!

", "rawMarkdown": "Congrats to you !!", "votes": 1 }, { "id": 479079, "postDate": "2019-02-27T00:48:35.520Z", "content": "

Congrats. trusting cv

", "rawMarkdown": "Congrats. trusting cv", "votes": 1 }, { "id": 479077, "postDate": "2019-02-27T00:46:31.307Z", "content": "

Congrats!!

\n\n

Love to hear more about your methods and the lessons learned.

", "rawMarkdown": "Congrats!!\n\nLove to hear more about your methods and the lessons learned.", "votes": 1 }, { "id": 479073, "postDate": "2019-02-27T00:43:45.290Z", "content": "

Congrats. Trust your CV.

", "rawMarkdown": "Congrats. Trust your CV.", "votes": 1 }, { "id": 479051, "postDate": "2019-02-27T00:23:43.480Z", "content": "

congrats! but trusting cv didnt work for me...

", "rawMarkdown": "congrats! but trusting cv didnt work for me...", "votes": 1, "replies": [ { "id": 479087, "postDate": "2019-02-27T00:53:13.107Z", "content": "

I think I conclude too fast about CV was winning, it can be I am too lucky. Let's see other people CVs.

", "rawMarkdown": "I think I conclude too fast about CV was winning, it can be I am too lucky. Let's see other people CVs.", "votes": 1 } ] }, { "id": 479044, "postDate": "2019-02-27T00:17:58.887Z", "content": "

Well done.

", "rawMarkdown": "Well done.", "votes": 1 }, { "id": 479041, "postDate": "2019-02-27T00:12:52.143Z", "content": "

Congrats! Yes, CV is king.

", "rawMarkdown": "Congrats! Yes, CV is king.", "votes": 1 }, { "id": 479037, "postDate": "2019-02-27T00:08:44.843Z", "content": "

Congrats. Trusting CV works this time.

", "rawMarkdown": "Congrats. Trusting CV works this time. ", "votes": 1 }, { "id": 479470, "postDate": "2019-02-27T07:04:58.513Z", "content": "

Congrats!

", "rawMarkdown": "Congrats! ", "votes": 2 }, { "id": 479156, "postDate": "2019-02-27T01:50:15.857Z", "content": "

Congratulation nlgn. CV is king, but what kind of strategy did you use ? Kfold ? Stratified using outliers as split ? I found this piece of code here where you reallocate the index of data sampling equally distributions according to target really cool:\n```\ntrain['rounded_target'] = train['target'].round(0)\n train = train.sort_values('rounded_target').reset_index(drop=True)

\n\n
vc = train['rounded_target'].value_counts()\nvc = dict(sorted(vc.items()))\ndf = pd.DataFrame()\ntrain['indexcol'] , i = 0,1\nfor k,v in vc.items():\n    step = train.shape[0]/v\n    indent = train.shape[0]/(v+1)\n    df2 = train[train['rounded_target'] == k].sample(v, random_state=120).reset_index(drop=True)\n    for j in range(0, v):\n        df2.at[j, 'indexcol'] = indent + j*step + 0.000001*i\n    df = pd.concat([df2,df])\n    i+=1\n\ntrain = df.sort_values('indexcol', ascending=True).reset_index(drop=True)\n\ndel train['indexcol'], train['rounded_target']\n
\n\n

```

", "rawMarkdown": "Congratulation nlgn. CV is king, but what kind of strategy did you use ? Kfold ? Stratified using outliers as split ? I found this piece of code here where you reallocate the index of data sampling equally distributions according to target really cool:\n```\ntrain['rounded_target'] = train['target'].round(0)\n train = train.sort_values('rounded_target').reset_index(drop=True)\n \n vc = train['rounded_target'].value_counts()\n vc = dict(sorted(vc.items()))\n df = pd.DataFrame()\n train['indexcol'] , i = 0,1\n for k,v in vc.items():\n step = train.shape[0]/v\n indent = train.shape[0]/(v+1)\n df2 = train[train['rounded_target'] == k].sample(v, random_state=120).reset_index(drop=True)\n for j in range(0, v):\n df2.at[j, 'indexcol'] = indent + j*step + 0.000001*i\n df = pd.concat([df2,df])\n i+=1\n \n train = df.sort_values('indexcol', ascending=True).reset_index(drop=True)\n \n del train['indexcol'], train['rounded_target']\n```", "votes": 2, "replies": [ { "id": 479486, "postDate": "2019-02-27T07:28:59.670Z", "content": "

I didn't find this script useful, I tried that and my cv went down. I used 5 fold stratified on outliers (0/1).

", "rawMarkdown": "I didn't find this script useful, I tried that and my cv went down. I used 5 fold stratified on outliers (0/1).", "votes": 1 }, { "id": 487264, "postDate": "2019-03-10T14:04:10.447Z", "content": "

Can you tell me where to see the winner's code ? I can't find that.

", "rawMarkdown": "Can you tell me where to see the winner's code ? I can't find that." } ] }, { "id": 479141, "postDate": "2019-02-27T01:39:11.810Z", "content": "

congrats, and great final subs selection.\nWe fail to select our top 12 subs@@

", "rawMarkdown": "congrats, and great final subs selection.\nWe fail to select our top 12 subs@@", "votes": 2, "replies": [ { "id": 479153, "postDate": "2019-02-27T01:48:29.497Z", "content": "

It is sad, your CVs look so good.

", "rawMarkdown": "It is sad, your CVs look so good." } ] }, { "id": 479132, "postDate": "2019-02-27T01:31:16.690Z", "content": "

It really just depends on the scale of the data. If your training set is much bigger than test set then I value CV much more. Sometimes that isnt the case though.

", "rawMarkdown": "It really just depends on the scale of the data. If your training set is much bigger than test set then I value CV much more. Sometimes that isnt the case though. ", "votes": 2 }, { "id": 479118, "postDate": "2019-02-27T01:18:14.937Z", "content": "

Congrats @kingychiu. Yes CV works especially with the best features.

\n\n

Update:

\n\n

Here is my post on feature selection techniques https://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82052

", "rawMarkdown": "Congrats @kingychiu. Yes CV works especially with the best features. \n\n**Update:**\n\nHere is my post on feature selection techniques https://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82052", "votes": 2 }, { "id": 932704, "postDate": "2020-07-17T08:16:17.707Z", "content": "

Congrats

", "rawMarkdown": "Congrats" }, { "id": 487253, "postDate": "2019-03-10T13:33:16.037Z", "content": "

Congratulations!

", "rawMarkdown": "Congratulations!" }, { "id": 487120, "postDate": "2019-03-10T07:07:15.087Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!" }, { "id": 486827, "postDate": "2019-03-09T13:53:34.950Z", "content": "

Well done....!

", "rawMarkdown": "Well done....!" }, { "id": 484748, "postDate": "2019-03-06T12:27:36.377Z", "content": "

Congratulations

", "rawMarkdown": "Congratulations" }, { "id": 484588, "postDate": "2019-03-06T07:52:40.803Z", "content": "

Congratulations!

", "rawMarkdown": "Congratulations!" }, { "id": 484483, "postDate": "2019-03-06T04:35:29.877Z", "content": "

Congrats

", "rawMarkdown": "Congrats" }, { "id": 483977, "postDate": "2019-03-05T12:08:23.937Z", "content": "

congrats!

", "rawMarkdown": "congrats!" }, { "id": 483889, "postDate": "2019-03-05T09:40:19.897Z", "content": "

Congratulations!!

", "rawMarkdown": "Congratulations!!\n" }, { "id": 483340, "postDate": "2019-03-04T13:29:01.813Z", "content": "

Congrats dude!

", "rawMarkdown": "Congrats dude!" }, { "id": 483336, "postDate": "2019-03-04T13:22:55.430Z", "content": "

Congratulation!!

", "rawMarkdown": "Congratulation!!" }, { "id": 483106, "postDate": "2019-03-04T06:29:43.717Z", "content": "

Congrats

", "rawMarkdown": "Congrats" }, { "id": 482528, "postDate": "2019-03-03T06:41:23.363Z", "content": "

Congratulations.

", "rawMarkdown": "Congratulations." }, { "id": 482194, "postDate": "2019-03-02T13:30:15.513Z", "content": "

congratulations!

", "rawMarkdown": "congratulations!" }, { "id": 482010, "postDate": "2019-03-02T07:29:42.090Z", "content": "

congrats

", "rawMarkdown": "congrats" }, { "id": 480881, "postDate": "2019-02-28T20:16:01.023Z", "content": "

Congrats

", "rawMarkdown": "Congrats" }, { "id": 480801, "postDate": "2019-02-28T17:29:12.380Z", "content": "

Congratulations :)

", "rawMarkdown": "Congratulations :)" }, { "id": 480784, "postDate": "2019-02-28T17:01:25.477Z", "content": "

Congratulations

", "rawMarkdown": "Congratulations" }, { "id": 480748, "postDate": "2019-02-28T16:06:31.197Z", "content": "

Congratulations. Interesting solution.

", "rawMarkdown": "Congratulations. Interesting solution." }, { "id": 480720, "postDate": "2019-02-28T15:20:47.133Z", "content": "

Congratulations!

", "rawMarkdown": "Congratulations!" }, { "id": 480684, "postDate": "2019-02-28T14:17:12.903Z", "content": "

congratulations!

", "rawMarkdown": "congratulations!" }, { "id": 480665, "postDate": "2019-02-28T13:59:58.723Z", "content": "

congratulations.

", "rawMarkdown": "congratulations." }, { "id": 480654, "postDate": "2019-02-28T13:46:16.543Z", "content": "

Congratulations for your first gold medal

", "rawMarkdown": "Congratulations for your first gold medal" }, { "id": 480546, "postDate": "2019-02-28T10:17:20.783Z", "content": "

Congrats

", "rawMarkdown": "Congrats" }, { "id": 480448, "postDate": "2019-02-28T07:49:26.630Z", "content": "

Congrats!! Good job!

", "rawMarkdown": "Congrats!! Good job!" }, { "id": 480346, "postDate": "2019-02-28T04:35:20.227Z", "content": "

congratulation

", "rawMarkdown": "congratulation" }, { "id": 480288, "postDate": "2019-02-28T02:30:58.603Z", "content": "

Congratulations on your first solo gold!

", "rawMarkdown": "Congratulations on your first solo gold! " }, { "id": 480238, "postDate": "2019-02-28T00:58:08.287Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!" }, { "id": 480234, "postDate": "2019-02-28T00:40:01.277Z", "content": "

Congrats!

", "rawMarkdown": "Congrats!" }, { "id": 480024, "postDate": "2019-02-27T17:34:12.243Z", "content": "

Congratulation. keep going

", "rawMarkdown": "Congratulation. keep going" }, { "id": 479988, "postDate": "2019-02-27T16:46:57.100Z", "content": "

congratulations

", "rawMarkdown": "congratulations" }, { "id": 479942, "postDate": "2019-02-27T15:46:25.050Z", "content": "

Congratulations

", "rawMarkdown": "Congratulations " }, { "id": 479928, "postDate": "2019-02-27T15:26:17.513Z", "content": "

congratz!

", "rawMarkdown": "congratz! " }, { "id": 479987, "postDate": "2019-02-27T16:46:00.380Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 480493, "author_name": "Sarmat Kasaev", "author_url": "", "post_date": "2019-02-28T08:40:09.360000", "content": "

Congratulations!

", "votes": 1, "replies": [] }, { "id": 479995, "author_name": "Can Korkut", "author_url": "", "post_date": "2019-02-27T16:56:26.120000", "content": "

Congratulations.

", "votes": 1, "replies": [] }, { "id": 479880, "author_name": "Gaurav Adhikari", "author_url": "", "post_date": "2019-02-27T14:29:56.410000", "content": "

congratulations (y)

", "votes": 1, "replies": [] }, { "id": 479795, "author_name": "FuengWul", "author_url": "", "post_date": "2019-02-27T12:35:27.553000", "content": "

Congratulations for your first gold medal

", "votes": 1, "replies": [] }, { "id": 479758, "author_name": "Shubham Mandowara", "author_url": "", "post_date": "2019-02-27T11:58:48.243000", "content": "

Congratulations.

", "votes": 1, "replies": [] }, { "id": 479737, "author_name": "Jie Wu", "author_url": "", "post_date": "2019-02-27T11:31:10.763000", "content": "

Congrats

", "votes": 1, "replies": [] }, { "id": 479734, "author_name": "Rahul Kalgunde", "author_url": "", "post_date": "2019-02-27T11:29:00.230000", "content": "

Congrats!!

", "votes": 1, "replies": [] }, { "id": 479636, "author_name": "Bala Subrahmanyam", "author_url": "", "post_date": "2019-02-27T09:56:56.910000", "content": "

Congratulations on your first win

", "votes": 1, "replies": [] }, { "id": 479576, "author_name": "SubikashPal", "author_url": "", "post_date": "2019-02-27T08:54:31.010000", "content": "

congratulations !!! well deserved.

", "votes": 1, "replies": [] }, { "id": 479573, "author_name": "kira0769", "author_url": "", "post_date": "2019-02-27T08:54:11.820000", "content": "

Congratulations

", "votes": 1, "replies": [] }, { "id": 479567, "author_name": "François", "author_url": "", "post_date": "2019-02-27T08:49:58.750000", "content": "

Congratulation ! I'm looking forward to see your work.

", "votes": 1, "replies": [] }, { "id": 479478, "author_name": "Karthik Chowdary Tsaliki", "author_url": "", "post_date": "2019-02-27T07:18:05.200000", "content": "

Congratulations @kingychiu Keep up the great work going.

", "votes": 1, "replies": [] }, { "id": 479450, "author_name": "Manchoi", "author_url": "", "post_date": "2019-02-27T06:36:24.653000", "content": "

Congratulations!!! and could you please share more about your solution, such as feature engineering, models, etc..

", "votes": 1, "replies": [] }, { "id": 479404, "author_name": "Icfstat", "author_url": "", "post_date": "2019-02-27T05:37:14.523000", "content": "

Congratulations!!! . Keep up your good work and you will get the grandmaster tier soon! . I was dreaming of getting a solo gold medal, I took high risk with post processing and I did not select my best single models as finals submissions which would have been at least very near to gold positions. This situation tells me that I still have to improve my skills and work harder to get that fantastic solo gold medal ! . Congratulations again !

", "votes": 1, "replies": [ { "id": 479993, "author_name": "Anthony Chiu", "author_url": "", "post_date": "2019-02-27T16:53:57.993000", "content": "

Thanks for your kind words. I was so scared about manual post-processing and haven't done it. We are all learning here :)

", "votes": 1, "replies": [] } ] }, { "id": 479369, "author_name": " Conggo", "author_url": "", "post_date": "2019-02-27T05:10:44.867000", "content": "

Congrats!

", "votes": 1, "replies": [] }, { "id": 479355, "author_name": "Vishy", "author_url": "", "post_date": "2019-02-27T04:58:15.560000", "content": "

Congratulations, Guess you have held on to you CV, Could you share how was CV selecting done by you like, number of rows in Local CV, Random rows or first/last n rows etc..

", "votes": 1, "replies": [ { "id": 479487, "author_name": "Anthony Chiu", "author_url": "", "post_date": "2019-02-27T07:30:09.200000", "content": "

Thanks, what do you mean by CV selecting? I simply picked the highest CV and the highest LB as the final 2 submissions.

", "votes": 1, "replies": [] } ] }, { "id": 479337, "author_name": "mrxnew", "author_url": "", "post_date": "2019-02-27T04:35:27.190000", "content": "

Congrats...

", "votes": 1, "replies": [] }, { "id": 479317, "author_name": "Rraj", "author_url": "", "post_date": "2019-02-27T04:14:14.140000", "content": "

Congratulations ! @nlgn.\nWe hope you would share your strategy and insights with the community. Would love to learn from your work.

", "votes": 1, "replies": [] }, { "id": 479312, "author_name": "Nooh", "author_url": "", "post_date": "2019-02-27T04:08:02.950000", "content": "

Congrats @nlgn. It was indeed a tricky competition. And I'm happy that I trusted my CV. CV FTW!

", "votes": 1, "replies": [] }, { "id": 479292, "author_name": "senkin13", "author_url": "", "post_date": "2019-02-27T03:49:00.007000", "content": "

Congrats!

", "votes": 1, "replies": [] }, { "id": 479284, "author_name": "good good study", "author_url": "", "post_date": "2019-02-27T03:38:16.573000", "content": "

Congrats! Trusting CV wins

", "votes": 1, "replies": [] }, { "id": 479233, "author_name": "xffj", "author_url": "", "post_date": "2019-02-27T03:05:01.553000", "content": "

say nothing more, congrats!👍

", "votes": 1, "replies": [] }, { "id": 479175, "author_name": "George", "author_url": "", "post_date": "2019-02-27T02:10:50.977000", "content": "

Congrats!

", "votes": 1, "replies": [] }, { "id": 479158, "author_name": "Zhuang Jia", "author_url": "", "post_date": "2019-02-27T01:51:02.433000", "content": "

very amazing, your blending is quite stable

", "votes": 1, "replies": [] }, { "id": 479143, "author_name": "zhbain", "author_url": "", "post_date": "2019-02-27T01:40:25.293000", "content": "

Congrats! Trust your CV is first rule, LOL.

", "votes": 1, "replies": [] }, { "id": 479138, "author_name": "Rajath", "author_url": "", "post_date": "2019-02-27T01:36:19.910000", "content": "

Congratulations. Would love to see an writeup of your solution.

", "votes": 1, "replies": [] }, { "id": 479136, "author_name": "LongYin/杰少", "author_url": "", "post_date": "2019-02-27T01:34:32.900000", "content": "

Congrats!

", "votes": 1, "replies": [] }, { "id": 479133, "author_name": "🐳鲲(China)", "author_url": "", "post_date": "2019-02-27T01:31:41.287000", "content": "

more is coming

", "votes": 1, "replies": [] }, { "id": 479120, "author_name": "Vishy", "author_url": "", "post_date": "2019-02-27T01:19:38.657000", "content": "

Congrats to you !!

", "votes": 1, "replies": [] }, { "id": 479079, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:48:35.520000", "content": "", "votes": 1, "replies": [] }, { "id": 479077, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:46:31.307000", "content": "", "votes": 1, "replies": [] }, { "id": 479073, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:43:45.290000", "content": "", "votes": 1, "replies": [] }, { "id": 479051, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:23:43.480000", "content": "", "votes": 1, "replies": [ { "id": 479087, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:53:13.107000", "content": "", "votes": 1, "replies": [] } ] }, { "id": 479044, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:17:58.887000", "content": "", "votes": 1, "replies": [] }, { "id": 479041, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:12:52.143000", "content": "", "votes": 1, "replies": [] }, { "id": 479037, "author_name": "", "author_url": "", "post_date": "2019-02-27T00:08:44.843000", "content": "", "votes": 1, "replies": [] }, { "id": 479470, "author_name": "", "author_url": "", "post_date": "2019-02-27T07:04:58.513000", "content": "", "votes": 2, "replies": [] }, { "id": 479156, "author_name": "", "author_url": "", "post_date": "2019-02-27T01:50:15.857000", "content": "", "votes": 2, "replies": [ { "id": 479486, "author_name": "", "author_url": "", "post_date": "2019-02-27T07:28:59.670000", "content": "", "votes": 1, "replies": [] }, { "id": 487264, "author_name": "", "author_url": "", "post_date": "2019-03-10T14:04:10.447000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 479141, "author_name": "", "author_url": "", "post_date": "2019-02-27T01:39:11.810000", "content": "", "votes": 2, "replies": [ { "id": 479153, "author_name": "", "author_url": "", "post_date": "2019-02-27T01:48:29.497000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 479132, "author_name": "", "author_url": "", "post_date": "2019-02-27T01:31:16.690000", "content": "", "votes": 2, "replies": [] }, { "id": 479118, "author_name": "", "author_url": "", "post_date": "2019-02-27T01:18:14.937000", "content": "", "votes": 2, "replies": [] }, { "id": 932704, "author_name": "", "author_url": "", "post_date": "2020-07-17T08:16:17.707000", "content": "", "votes": 0, "replies": [] }, { "id": 487253, "author_name": "", "author_url": "", "post_date": "2019-03-10T13:33:16.037000", "content": "", "votes": 0, "replies": [] }, { "id": 487120, "author_name": "", "author_url": "", "post_date": "2019-03-10T07:07:15.087000", "content": "", "votes": 0, "replies": [] }, { "id": 486827, "author_name": "", "author_url": "", "post_date": "2019-03-09T13:53:34.950000", "content": "", "votes": 0, "replies": [] }, { "id": 484748, "author_name": "", "author_url": "", "post_date": "2019-03-06T12:27:36.377000", "content": "", "votes": 0, "replies": [] }, { "id": 484588, "author_name": "", "author_url": "", "post_date": "2019-03-06T07:52:40.803000", "content": "", "votes": 0, "replies": [] }, { "id": 484483, "author_name": "", "author_url": "", "post_date": "2019-03-06T04:35:29.877000", "content": "", "votes": 0, "replies": [] }, { "id": 483977, "author_name": "", "author_url": "", "post_date": "2019-03-05T12:08:23.937000", "content": "", "votes": 0, "replies": [] }, { "id": 483889, "author_name": "", "author_url": "", "post_date": "2019-03-05T09:40:19.897000", "content": "", "votes": 0, "replies": [] }, { "id": 483340, "author_name": "", "author_url": "", "post_date": "2019-03-04T13:29:01.813000", "content": "", "votes": 0, "replies": [] }, { "id": 483336, "author_name": "", "author_url": "", "post_date": "2019-03-04T13:22:55.430000", "content": "", "votes": 0, "replies": [] }, { "id": 483106, "author_name": "", "author_url": "", "post_date": "2019-03-04T06:29:43.717000", "content": "", "votes": 0, "replies": [] }, { "id": 482528, "author_name": "", "author_url": "", "post_date": "2019-03-03T06:41:23.363000", "content": "", "votes": 0, "replies": [] }, { "id": 482194, "author_name": "", "author_url": "", "post_date": "2019-03-02T13:30:15.513000", "content": "", "votes": 0, "replies": [] }, { "id": 482010, "author_name": "", "author_url": "", "post_date": "2019-03-02T07:29:42.090000", "content": "", "votes": 0, "replies": [] }, { "id": 480881, "author_name": "", "author_url": "", "post_date": "2019-02-28T20:16:01.023000", "content": "", "votes": 0, "replies": [] }, { "id": 480801, "author_name": "", "author_url": "", "post_date": "2019-02-28T17:29:12.380000", "content": "", "votes": 0, "replies": [] }, { "id": 480784, "author_name": "", "author_url": "", "post_date": "2019-02-28T17:01:25.477000", "content": "", "votes": 0, "replies": [] }, { "id": 480748, "author_name": "", "author_url": "", "post_date": "2019-02-28T16:06:31.197000", "content": "", "votes": 0, "replies": [] }, { "id": 480720, "author_name": "", "author_url": "", "post_date": "2019-02-28T15:20:47.133000", "content": "", "votes": 0, "replies": [] }, { "id": 480684, "author_name": "", "author_url": "", "post_date": "2019-02-28T14:17:12.903000", "content": "", "votes": 0, "replies": [] }, { "id": 480665, "author_name": "", "author_url": "", "post_date": "2019-02-28T13:59:58.723000", "content": "", "votes": 0, "replies": [] }, { "id": 480654, "author_name": "", "author_url": "", "post_date": "2019-02-28T13:46:16.543000", "content": "", "votes": 0, "replies": [] }, { "id": 480546, "author_name": "", "author_url": "", "post_date": "2019-02-28T10:17:20.783000", "content": "", "votes": 0, "replies": [] }, { "id": 480448, "author_name": "", "author_url": "", "post_date": "2019-02-28T07:49:26.630000", "content": "", "votes": 0, "replies": [] }, { "id": 480346, "author_name": "", "author_url": "", "post_date": "2019-02-28T04:35:20.227000", "content": "", "votes": 0, "replies": [] }, { "id": 480288, "author_name": "", "author_url": "", "post_date": "2019-02-28T02:30:58.603000", "content": "", "votes": 0, "replies": [] }, { "id": 480238, "author_name": "", "author_url": "", "post_date": "2019-02-28T00:58:08.287000", "content": "", "votes": 0, "replies": [] }, { "id": 480234, "author_name": "", "author_url": "", "post_date": "2019-02-28T00:40:01.277000", "content": "", "votes": 0, "replies": [] }, { "id": 480024, "author_name": "", "author_url": "", "post_date": "2019-02-27T17:34:12.243000", "content": "", "votes": 0, "replies": [] }, { "id": 479988, "author_name": "", "author_url": "", "post_date": "2019-02-27T16:46:57.100000", "content": "", "votes": 0, "replies": [] }, { "id": 479942, "author_name": "", "author_url": "", "post_date": "2019-02-27T15:46:25.050000", "content": "", "votes": 0, "replies": [] }, { "id": 479928, "author_name": "", "author_url": "", "post_date": "2019-02-27T15:26:17.513000", "content": "", "votes": 0, "replies": [] }, { "id": 479987, "author_name": "", "author_url": "", "post_date": "2019-02-27T16:46:00.380000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "479034": "I was expecting a bronze and wishing for a silver. I was shocked in the morning.\nThanks all discussions and kernels here. It was a frustrating game. I have learned a lot here. \nDetails of what I did: \nhttps://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82166\n\nI think CV wins in this game right?\n\n![enter image description here][1]\n\n\n [1]: https://i.imgur.com/PM7UxW3.png\n", "480493": "Congratulations!", "479995": "Congratulations.", "479880": "congratulations (y)", "479795": "Congratulations for your first gold medal", "479758": "Congratulations.", "479737": "Congrats", "479734": "Congrats!!", "479636": "Congratulations on your first win ", "479576": "congratulations !!! well deserved.", "479573": "Congratulations", "479567": "Congratulation ! I'm looking forward to see your work. ", "479478": "Congratulations @kingychiu Keep up the great work going.", "479450": "Congratulations!!! and could you please share more about your solution, such as feature engineering, models, etc..", "479404": "Congratulations!!! . Keep up your good work and you will get the grandmaster tier soon! . I was dreaming of getting a solo gold medal, I took high risk with post processing and I did not select my best single models as finals submissions which would have been at least very near to gold positions. This situation tells me that I still have to improve my skills and work harder to get that fantastic solo gold medal ! . Congratulations again !", "479369": "Congrats!\n\n", "479355": "Congratulations, Guess you have held on to you CV, Could you share how was CV selecting done by you like, number of rows in Local CV, Random rows or first/last n rows etc..", "479337": "Congrats...", "479317": "Congratulations ! @nlgn.\nWe hope you would share your strategy and insights with the community. Would love to learn from your work. ", "479312": "Congrats @nlgn. It was indeed a tricky competition. And I'm happy that I trusted my CV. CV FTW!", "479292": "Congrats! ", "479284": "Congrats! Trusting CV wins", "479233": "say nothing more, congrats!👍", "479175": "Congrats!", "479158": "very amazing, your blending is quite stable", "479143": "Congrats! Trust your CV is first rule, LOL.", "479138": "Congratulations. Would love to see an writeup of your solution.", "479136": "Congrats!", "479133": "more is coming", "479120": "Congrats to you !!", "479079": "Congrats. trusting cv", "479077": "Congrats!!\n\nLove to hear more about your methods and the lessons learned.", "479073": "Congrats. Trust your CV.", "479051": "congrats! but trusting cv didnt work for me...", "479044": "Well done.", "479041": "Congrats! Yes, CV is king.", "479037": "Congrats. Trusting CV works this time. ", "479470": "Congrats! ", "479156": "Congratulation nlgn. CV is king, but what kind of strategy did you use ? Kfold ? Stratified using outliers as split ? I found this piece of code here where you reallocate the index of data sampling equally distributions according to target really cool:\n```\ntrain['rounded_target'] = train['target'].round(0)\n train = train.sort_values('rounded_target').reset_index(drop=True)\n \n vc = train['rounded_target'].value_counts()\n vc = dict(sorted(vc.items()))\n df = pd.DataFrame()\n train['indexcol'] , i = 0,1\n for k,v in vc.items():\n step = train.shape[0]/v\n indent = train.shape[0]/(v+1)\n df2 = train[train['rounded_target'] == k].sample(v, random_state=120).reset_index(drop=True)\n for j in range(0, v):\n df2.at[j, 'indexcol'] = indent + j*step + 0.000001*i\n df = pd.concat([df2,df])\n i+=1\n \n train = df.sort_values('indexcol', ascending=True).reset_index(drop=True)\n \n del train['indexcol'], train['rounded_target']\n```", "479141": "congrats, and great final subs selection.\nWe fail to select our top 12 subs@@", "479132": "It really just depends on the scale of the data. If your training set is much bigger than test set then I value CV much more. Sometimes that isnt the case though. ", "479118": "Congrats @kingychiu. Yes CV works especially with the best features. \n\n**Update:**\n\nHere is my post on feature selection techniques https://www.kaggle.com/c/elo-merchant-category-recommendation/discussion/82052", "932704": "Congrats", "487253": "Congratulations!", "487120": "Congrats!", "486827": "Well done....!", "484748": "Congratulations", "484588": "Congratulations!", "484483": "Congrats", "483977": "congrats!", "483889": "Congratulations!!\n", "483340": "Congrats dude!", "483336": "Congratulation!!", "483106": "Congrats", "482528": "Congratulations.", "482194": "congratulations!", "482010": "congrats", "480881": "Congrats", "480801": "Congratulations :)", "480784": "Congratulations", "480748": "Congratulations. Interesting solution.", "480720": "Congratulations!", "480684": "congratulations!", "480665": "congratulations.", "480654": "Congratulations for your first gold medal", "480546": "Congrats", "480448": "Congrats!! Good job!", "480346": "congratulation", "480288": "Congratulations on your first solo gold! ", "480238": "Congrats!", "480234": "Congrats!", "480024": "Congratulation. keep going", "479988": "congratulations", "479942": "Congratulations ", "479928": "congratz! ", "479987": "" } }