{
  "id": 79665,
  "title": "Looking for joining a team(Single model:CV  3.6659 , LB: 3.678)",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/79665",
  "author_name": "",
  "post_date": "2019-02-06T12:08:54.138943500Z",
  "votes": null,
  "comment_count": 9,
  "views": 0,
  "content": "<p>My current best model:\n5-fold LGB : CV  3.6659 , LB: 3.678</p>\n\n<p>there are few other LGB models in the range of LB:3.681 and 3.682.</p>\n\n<p>with post processing ( outlier detection and setting values properly) , I am able to achieve LB:3.667. Didn't spend too much time here.</p>\n\n<p>I am looking for some one who has good feature selection algorithm and good single model such that stacking may help us achieving good score.</p>\n\n<p>Please let me know if anyone interested to team up. </p>",
  "messages": [
    {
      "id": "467070",
      "postDate": "02/06/2019 12:08:54",
      "content": "<p>My current best model:\n5-fold LGB : CV  3.6659 , LB: 3.678</p>\n\n<p>there are few other LGB models in the range of LB:3.681 and 3.682.</p>\n\n<p>with post processing ( outlier detection and setting values properly) , I am able to achieve LB:3.667. Didn't spend too much time here.</p>\n\n<p>I am looking for some one who has good feature selection algorithm and good single model such that stacking may help us achieving good score.</p>\n\n<p>Please let me know if anyone interested to team up. </p>",
      "rawMarkdown": "My current best model:\n5-fold LGB : CV  3.6659 , LB: 3.678\n\nthere are few other LGB models in the range of LB:3.681 and 3.682.\n\nwith post processing ( outlier detection and setting values properly) , I am able to achieve LB:3.667. Didn't spend too much time here.\n\nI am looking for some one who has good feature selection algorithm and good single model such that stacking may help us achieving good score.\n\nPlease let me know if anyone interested to team up.",
      "votes": null
    },
    {
      "id": "467148",
      "postDate": "02/06/2019 14:45:24",
      "content": "<p>Hi @SubikashPal,\nYou already know from previous posts that  i've been working much on the stacking and ensembling side of things, creating diverse models (from 3.695 single model  to 3.679 LB no post processing)</p>\n\n<p>I also have prepared a few feed forward, backward, and permutation importance feature selection algorithms set up that we can use.</p>\n\n<p>I had an interest to create my own codebase for feature selection and blending, so that's what i worked on very extensively in this competition,  </p>",
      "rawMarkdown": "Hi @SubikashPal,\nYou already know from previous posts that  i've been working much on the stacking and ensembling side of things, creating diverse models (from 3.695 single model  to 3.679 LB no post processing)\n\nI also have prepared a few feed forward, backward, and permutation importance feature selection algorithms set up that we can use.\n\nI had an interest to create my own codebase for feature selection and blending, so that's what i worked on very extensively in this competition,",
      "votes": null
    },
    {
      "id": "467355",
      "postDate": "02/06/2019 23:39:27",
      "content": "<p>I’m willing to team up. Though I can’t say I have a whole lot to offer besides my plucky attitude and a decent desktop. </p>",
      "rawMarkdown": "I’m willing to team up. Though I can’t say I have a whole lot to offer besides my plucky attitude and a decent desktop.",
      "votes": null
    },
    {
      "id": "467469",
      "postDate": "02/07/2019 07:27:17",
      "content": "<p><a href=\"/interneuron\">@interneuron</a> What's your single model CV and LB?</p>",
      "rawMarkdown": "interneuron What's your single model CV and LB?",
      "votes": null
    },
    {
      "id": "467493",
      "postDate": "02/07/2019 08:15:59",
      "content": "<p>My best single model is 3.89 LB and around 3.64 with 7-fold CV.</p>",
      "rawMarkdown": "My best single model is 3.89 LB and around 3.64 with 7-fold CV.",
      "votes": null
    },
    {
      "id": "467514",
      "postDate": "02/07/2019 08:57:52",
      "content": "<p>Hi <a href=\"/interneuron\">@interneuron</a>,\nLet's wait a few days and see what happens.\nTo make a significant jump we need a stronger single model, or specifically its features.\nI'm still working on different stacking and diversification solutions.\nLet's contact again in a few days.</p>\n\n<p>Keep up the good work!!</p>",
      "rawMarkdown": "Hi @interneuron,\nLet's wait a few days and see what happens.\nTo make a significant jump we need a stronger single model, or specifically its features.\nI'm still working on different stacking and diversification solutions.\nLet's contact again in a few days.\n\nKeep up the good work!!",
      "votes": null
    },
    {
      "id": "467521",
      "postDate": "02/07/2019 09:07:16",
      "content": "<p>That's all for sure! Thanks for the response and no worries. \nBest of luck!</p>",
      "rawMarkdown": "That's all for sure! Thanks for the response and no worries. \nBest of luck!",
      "votes": null
    },
    {
      "id": "472457",
      "postDate": "02/16/2019 00:49:36",
      "content": "<p><a href=\"/interneuron\">@interneuron</a> are you still looking for a team? </p>",
      "rawMarkdown": "interneuron are you still looking for a team?",
      "votes": null
    },
    {
      "id": "472835",
      "postDate": "02/16/2019 18:25:19",
      "content": "<p>Hi, yes I am.</p>",
      "rawMarkdown": "Hi, yes I am.",
      "votes": null
    },
    {
      "id": "473161",
      "postDate": "02/17/2019 13:02:38",
      "content": "<p><a href=\"/interneuron\">@interneuron</a> would you be interested in working together? I've sent you a kaggle message. </p>",
      "rawMarkdown": "interneuron would you be interested in working together? I've sent you a kaggle message.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 467148,
      "author_name": "tpthegreat",
      "author_url": "",
      "post_date": "02/06/2019 14:45:24",
      "content": "<p>Hi @SubikashPal,\nYou already know from previous posts that  i've been working much on the stacking and ensembling side of things, creating diverse models (from 3.695 single model  to 3.679 LB no post processing)</p>\n\n<p>I also have prepared a few feed forward, backward, and permutation importance feature selection algorithms set up that we can use.</p>\n\n<p>I had an interest to create my own codebase for feature selection and blending, so that's what i worked on very extensively in this competition,  </p>",
      "votes": null,
      "replies": [
        {
          "id": 467355,
          "author_name": "interneuron",
          "author_url": "",
          "post_date": "02/06/2019 23:39:27",
          "content": "<p>I’m willing to team up. Though I can’t say I have a whole lot to offer besides my plucky attitude and a decent desktop. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 467469,
          "author_name": "tpthegreat",
          "author_url": "",
          "post_date": "02/07/2019 07:27:17",
          "content": "<p><a href=\"/interneuron\">@interneuron</a> What's your single model CV and LB?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 467493,
          "author_name": "interneuron",
          "author_url": "",
          "post_date": "02/07/2019 08:15:59",
          "content": "<p>My best single model is 3.89 LB and around 3.64 with 7-fold CV.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 467514,
          "author_name": "tpthegreat",
          "author_url": "",
          "post_date": "02/07/2019 08:57:52",
          "content": "<p>Hi <a href=\"/interneuron\">@interneuron</a>,\nLet's wait a few days and see what happens.\nTo make a significant jump we need a stronger single model, or specifically its features.\nI'm still working on different stacking and diversification solutions.\nLet's contact again in a few days.</p>\n\n<p>Keep up the good work!!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 467521,
          "author_name": "interneuron",
          "author_url": "",
          "post_date": "02/07/2019 09:07:16",
          "content": "<p>That's all for sure! Thanks for the response and no worries. \nBest of luck!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 472457,
          "author_name": "gideonteo",
          "author_url": "",
          "post_date": "02/16/2019 00:49:36",
          "content": "<p><a href=\"/interneuron\">@interneuron</a> are you still looking for a team? </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 472835,
          "author_name": "interneuron",
          "author_url": "",
          "post_date": "02/16/2019 18:25:19",
          "content": "<p>Hi, yes I am.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 473161,
          "author_name": "gideonteo",
          "author_url": "",
          "post_date": "02/17/2019 13:02:38",
          "content": "<p><a href=\"/interneuron\">@interneuron</a> would you be interested in working together? I've sent you a kaggle message. </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "467070": "My current best model:\n5-fold LGB : CV  3.6659 , LB: 3.678\n\nthere are few other LGB models in the range of LB:3.681 and 3.682.\n\nwith post processing ( outlier detection and setting values properly) , I am able to achieve LB:3.667. Didn't spend too much time here.\n\nI am looking for some one who has good feature selection algorithm and good single model such that stacking may help us achieving good score.\n\nPlease let me know if anyone interested to team up.",
    "467148": "Hi @SubikashPal,\nYou already know from previous posts that  i've been working much on the stacking and ensembling side of things, creating diverse models (from 3.695 single model  to 3.679 LB no post processing)\n\nI also have prepared a few feed forward, backward, and permutation importance feature selection algorithms set up that we can use.\n\nI had an interest to create my own codebase for feature selection and blending, so that's what i worked on very extensively in this competition,",
    "467355": "I’m willing to team up. Though I can’t say I have a whole lot to offer besides my plucky attitude and a decent desktop.",
    "467469": "interneuron What's your single model CV and LB?",
    "467493": "My best single model is 3.89 LB and around 3.64 with 7-fold CV.",
    "467514": "Hi @interneuron,\nLet's wait a few days and see what happens.\nTo make a significant jump we need a stronger single model, or specifically its features.\nI'm still working on different stacking and diversification solutions.\nLet's contact again in a few days.\n\nKeep up the good work!!",
    "467521": "That's all for sure! Thanks for the response and no worries. \nBest of luck!",
    "472457": "interneuron are you still looking for a team?",
    "472835": "Hi, yes I am.",
    "473161": "interneuron would you be interested in working together? I've sent you a kaggle message."
  },
  "source": "meta"
}