{
  "id": 175274,
  "title": "MinMax or np.log ensemble",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/175274",
  "author_name": "",
  "post_date": "2020-08-17T18:10:17.900255500Z",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hello everyone, right now I am using np.log ensemble and I have 2 submission left. I wanted to know<br>\nis MinMax ensemble worth trying?</p>",
  "messages": [
    {
      "id": "974094",
      "postDate": "08/17/2020 18:10:17",
      "content": "<p>Hello everyone, right now I am using np.log ensemble and I have 2 submission left. I wanted to know<br>\nis MinMax ensemble worth trying?</p>",
      "rawMarkdown": "Hello everyone, right now I am using np.log ensemble and I have 2 submission left. I wanted to know\nis MinMax ensemble worth trying?",
      "votes": null
    },
    {
      "id": "974141",
      "postDate": "08/17/2020 18:38:43",
      "content": "<p>MinMax didn't work in my case. Log averaging worked better.</p>",
      "rawMarkdown": "MinMax didn't work in my case. Log averaging worked better.",
      "votes": null
    },
    {
      "id": "974227",
      "postDate": "08/17/2020 20:10:15",
      "content": "<p>Yes, if you can set a good low and high threshold. Which you can do by playing around with your validation set.</p>",
      "rawMarkdown": "Yes, if you can set a good low and high threshold. Which you can do by playing around with your validation set.",
      "votes": null
    },
    {
      "id": "974239",
      "postDate": "08/17/2020 20:23:27",
      "content": "<p>It perfectly works at the public dataset but often causes low scores on the private one. So you can easily get a good score with that but you should be ready for a shake-up.</p>",
      "rawMarkdown": "It perfectly works at the public dataset but often causes low scores on the private one. So you can easily get a good score with that but you should be ready for a shake-up.",
      "votes": null
    },
    {
      "id": "974244",
      "postDate": "08/17/2020 20:29:00",
      "content": "<p>Which one of these you are using.</p>",
      "rawMarkdown": "Which one of these you are using.",
      "votes": null
    },
    {
      "id": "974265",
      "postDate": "08/17/2020 20:55:01",
      "content": "<p>I haven't used MinMax. It's simple overfitting, I don't see any reasons to spend any submits for this. However, there 3 submissions for the final score, so you can try your luck.</p>",
      "rawMarkdown": "I haven't used MinMax. It's simple overfitting, I don't see any reasons to spend any submits for this. However, there 3 submissions for the final score, so you can try your luck.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 974141,
      "author_name": "prateek0x",
      "author_url": "",
      "post_date": "08/17/2020 18:38:43",
      "content": "<p>MinMax didn't work in my case. Log averaging worked better.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 974239,
      "author_name": "vadimtimakin",
      "author_url": "",
      "post_date": "08/17/2020 20:23:27",
      "content": "<p>It perfectly works at the public dataset but often causes low scores on the private one. So you can easily get a good score with that but you should be ready for a shake-up.</p>",
      "votes": null,
      "replies": [
        {
          "id": 974244,
          "author_name": "pramod722445",
          "author_url": "",
          "post_date": "08/17/2020 20:29:00",
          "content": "<p>Which one of these you are using.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 974265,
          "author_name": "vadimtimakin",
          "author_url": "",
          "post_date": "08/17/2020 20:55:01",
          "content": "<p>I haven't used MinMax. It's simple overfitting, I don't see any reasons to spend any submits for this. However, there 3 submissions for the final score, so you can try your luck.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 974227,
      "author_name": "vikrant06",
      "author_url": "",
      "post_date": "08/17/2020 20:10:15",
      "content": "<p>Yes, if you can set a good low and high threshold. Which you can do by playing around with your validation set.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "974094": "Hello everyone, right now I am using np.log ensemble and I have 2 submission left. I wanted to know\nis MinMax ensemble worth trying?",
    "974141": "MinMax didn't work in my case. Log averaging worked better.",
    "974227": "Yes, if you can set a good low and high threshold. Which you can do by playing around with your validation set.",
    "974239": "It perfectly works at the public dataset but often causes low scores on the private one. So you can easily get a good score with that but you should be ready for a shake-up.",
    "974244": "Which one of these you are using.",
    "974265": "I haven't used MinMax. It's simple overfitting, I don't see any reasons to spend any submits for this. However, there 3 submissions for the final score, so you can try your luck."
  },
  "source": "meta"
}