{
  "id": 29001,
  "title": "Takeaway for R users - 0.68 on laptop- FTRL, data.table and feature hashing",
  "url": "/competitions/outbrain-click-prediction/discussion/29001",
  "author_name": "diaman",
  "post_date": "2017-02-19T12:11:09.062000",
  "votes": 12,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi all.  As part of our solution we used FTRL. Here I summarized  how to achieve ~50 place with R (and little bit Rcpp) on a laptop:</p>\n\n<ol>\n<li><a href=\"https://dsnotes.com/post/2017-01-27-lessons-learned-from-outbrain-click-prediction-kaggle-competition/\">Part 1 - Large data, feature hashing and online learning</a></li>\n<li><a href=\"https://dsnotes.com/post/2017-02-07-large-data-feature-hashing-and-online-learning-part-2/\">Part 2 - Fitting logistic regression on 100gb dataset on a laptop</a></li>\n</ol>\n\n<p>EDIT - repo with easily reproducible end-to-end code <a href=\"https://github.com/dselivanov/kaggle-outbrain\">https://github.com/dselivanov/kaggle-outbrain</a></p>",
  "messages": [
    {
      "id": 162486,
      "postDate": "2017-02-19T12:11:09.063Z",
      "content": "<p>Hi all.  As part of our solution we used FTRL. Here I summarized  how to achieve ~50 place with R (and little bit Rcpp) on a laptop:</p>\n\n<ol>\n<li><a href=\"https://dsnotes.com/post/2017-01-27-lessons-learned-from-outbrain-click-prediction-kaggle-competition/\">Part 1 - Large data, feature hashing and online learning</a></li>\n<li><a href=\"https://dsnotes.com/post/2017-02-07-large-data-feature-hashing-and-online-learning-part-2/\">Part 2 - Fitting logistic regression on 100gb dataset on a laptop</a></li>\n</ol>\n\n<p>EDIT - repo with easily reproducible end-to-end code <a href=\"https://github.com/dselivanov/kaggle-outbrain\">https://github.com/dselivanov/kaggle-outbrain</a></p>",
      "rawMarkdown": "Hi all.  As part of our solution we used FTRL. Here I summarized  how to achieve ~50 place with R (and little bit Rcpp) on a laptop:\n\n1. [Part 1 - Large data, feature hashing and online learning](https://dsnotes.com/post/2017-01-27-lessons-learned-from-outbrain-click-prediction-kaggle-competition/)\n1. [Part 2 - Fitting logistic regression on 100gb dataset on a laptop](https://dsnotes.com/post/2017-02-07-large-data-feature-hashing-and-online-learning-part-2/)\n\nEDIT - repo with easily reproducible end-to-end code https://github.com/dselivanov/kaggle-outbrain",
      "votes": 12
    },
    {
      "id": 337172,
      "postDate": "2018-06-02T06:33:26.257Z",
      "content": "<p>@Pengju fixed - try again</p>",
      "rawMarkdown": "@Pengju fixed - try again",
      "replies": [
        {
          "id": 337177,
          "postDate": "2018-06-02T06:46:45Z",
          "content": "<p>It worked. tks for your kind help.</p>",
          "rawMarkdown": "It worked. tks for your kind help."
        }
      ]
    },
    {
      "id": 337166,
      "postDate": "2018-06-02T06:11:24.983Z",
      "content": "<p>hi diaman, both links are no longer available. could you pls. post the new links or send me by email? tks in advance. \nmy email zhaopengju1988@163.com </p>",
      "rawMarkdown": " hi diaman, both links are no longer available. could you pls. post the new links or send me by email? tks in advance. \nmy email zhaopengju1988@163.com "
    },
    {
      "id": 326191,
      "postDate": "2018-05-09T12:34:27.357Z",
      "content": "<p>linnks Part 1 - Large data, feature hashing and online learning\nPart 2 - Fitting logistic regression on 100gb dataset on a laptop \nare broken </p>",
      "rawMarkdown": "linnks Part 1 - Large data, feature hashing and online learning\nPart 2 - Fitting logistic regression on 100gb dataset on a laptop \nare broken "
    },
    {
      "id": 433561,
      "postDate": "2018-12-05T07:23:32.090Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 166732,
      "postDate": "2017-03-10T19:56:27.320Z",
      "content": "<p>Thanks!</p>",
      "rawMarkdown": "Thanks!"
    }
  ],
  "comments": [
    {
      "id": 337172,
      "author_name": "diaman",
      "author_url": "",
      "post_date": "2018-06-02T06:33:26.257000",
      "content": "<p>@Pengju fixed - try again</p>",
      "votes": 0,
      "replies": [
        {
          "id": 337177,
          "author_name": "Pengju",
          "author_url": "",
          "post_date": "2018-06-02T06:46:45",
          "content": "<p>It worked. tks for your kind help.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 337166,
      "author_name": "Pengju",
      "author_url": "",
      "post_date": "2018-06-02T06:11:24.983000",
      "content": "<p>hi diaman, both links are no longer available. could you pls. post the new links or send me by email? tks in advance. \nmy email zhaopengju1988@163.com </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 326191,
      "author_name": "Sandy",
      "author_url": "",
      "post_date": "2018-05-09T12:34:27.357000",
      "content": "<p>linnks Part 1 - Large data, feature hashing and online learning\nPart 2 - Fitting logistic regression on 100gb dataset on a laptop \nare broken </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 433561,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-12-05T07:23:32.090000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 166732,
      "author_name": "Nell",
      "author_url": "",
      "post_date": "2017-03-10T19:56:27.320000",
      "content": "<p>Thanks!</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "162486": "Hi all.  As part of our solution we used FTRL. Here I summarized  how to achieve ~50 place with R (and little bit Rcpp) on a laptop:\n\n1. [Part 1 - Large data, feature hashing and online learning](https://dsnotes.com/post/2017-01-27-lessons-learned-from-outbrain-click-prediction-kaggle-competition/)\n1. [Part 2 - Fitting logistic regression on 100gb dataset on a laptop](https://dsnotes.com/post/2017-02-07-large-data-feature-hashing-and-online-learning-part-2/)\n\nEDIT - repo with easily reproducible end-to-end code https://github.com/dselivanov/kaggle-outbrain",
    "337172": "@Pengju fixed - try again",
    "337166": " hi diaman, both links are no longer available. could you pls. post the new links or send me by email? tks in advance. \nmy email zhaopengju1988@163.com ",
    "326191": "linnks Part 1 - Large data, feature hashing and online learning\nPart 2 - Fitting logistic regression on 100gb dataset on a laptop \nare broken ",
    "433561": "",
    "166732": "Thanks!"
  }
}