{
  "id": 8322,
  "title": "Other Base Learners",
  "url": "/competitions/acquire-valued-shoppers-challenge/discussion/8322",
  "author_name": "",
  "post_date": "2014-05-30T07:09:53.950Z",
  "votes": null,
  "comment_count": 3,
  "views": 1296,
  "content": "<p>In the current posts, I noticed people are using Logistic regression, VW etc... Is anyone using other base learners like Random Forests (RF), Neural network (NN), SVM ....</p>\n<p>And what sort of performance/results are you achieving. I have been using RF so far given its bootstrap capability which suits a large dataset problem like this. Still early days for me but would be interested to hear what others are using.</p>\n\n<p>Thanks.</p>",
  "messages": [
    {
      "id": "47311",
      "postDate": "05/30/2014 07:09:53",
      "content": "<p>In the current posts, I noticed people are using Logistic regression, VW etc... Is anyone using other base learners like Random Forests (RF), Neural network (NN), SVM ....</p>\n<p>And what sort of performance/results are you achieving. I have been using RF so far given its bootstrap capability which suits a large dataset problem like this. Still early days for me but would be interested to hear what others are using.</p>\n\n<p>Thanks.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "47314",
      "postDate": "05/30/2014 08:16:11",
      "content": "<p>Yes, we use other base learner (with parameters set for very rough model tuning) -- not logistic regression and not VW, but we plan to&nbsp;try VW in the future.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "47330",
      "postDate": "05/30/2014 11:38:54",
      "content": "<p>If it is possible, what sort of performance ( just rough) figures you are achieving with those models?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "47332",
      "postDate": "05/30/2014 12:35:06",
      "content": "<p>I think base learner is not the point in this task, we&nbsp;use other methodology, other factors, other base learner -- it very hard to compare it all in once. And we will try to modify all of this (especially base learner)</p>\n<p>If you want to know what is our solutions maximum score with this base learner -- it is in leaderboard <strong>0.62131</strong>.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 47314,
      "author_name": "dremovd",
      "author_url": "",
      "post_date": "05/30/2014 08:16:11",
      "content": "<p>Yes, we use other base learner (with parameters set for very rough model tuning) -- not logistic regression and not VW, but we plan to&nbsp;try VW in the future.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 47330,
      "author_name": "upulbandara",
      "author_url": "",
      "post_date": "05/30/2014 11:38:54",
      "content": "<p>If it is possible, what sort of performance ( just rough) figures you are achieving with those models?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 47332,
      "author_name": "dremovd",
      "author_url": "",
      "post_date": "05/30/2014 12:35:06",
      "content": "<p>I think base learner is not the point in this task, we&nbsp;use other methodology, other factors, other base learner -- it very hard to compare it all in once. And we will try to modify all of this (especially base learner)</p>\n<p>If you want to know what is our solutions maximum score with this base learner -- it is in leaderboard <strong>0.62131</strong>.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "47311": "",
    "47314": "",
    "47330": "",
    "47332": ""
  },
  "source": "meta"
}