{
  "id": 219560,
  "title": "[Resolved] mean of predictions-submission?",
  "url": "/competitions/rfcx-species-audio-detection/discussion/219560",
  "author_name": "",
  "post_date": "2021-02-15T12:21:56.915665200Z",
  "votes": 5,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hello!!</p>\n<p>I am calculating the probability for each second, now for a clip of 60 sec, there will be 60 probabilities</p>\n<pre><code>my 1 probability= probability of all 24 species for a 1 sec clip\n</code></pre>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107949494-5a54b380-6fbb-11eb-9027-c030673ccf86.png\" alt=\"Screenshot 2021-02-15 at 6 26 12 PM\"></p>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107950221-6e4ce500-6fbc-11eb-9900-91241aa73451.png\" alt=\"Screenshot 2021-02-15 at 6 34 00 PM\"></p>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107952170-693d6500-6fbf-11eb-8a8c-e93b54c0f693.png\" alt=\"Screenshot 2021-02-15 at 6 55 14 PM\"></p>\n<p>Thanks,<br>\nArindam</p>",
  "messages": [
    {
      "id": "1201472",
      "postDate": "02/15/2021 12:21:56",
      "content": "<p>Hello!!</p>\n<p>I am calculating the probability for each second, now for a clip of 60 sec, there will be 60 probabilities</p>\n<pre><code>my 1 probability= probability of all 24 species for a 1 sec clip\n</code></pre>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107949494-5a54b380-6fbb-11eb-9027-c030673ccf86.png\" alt=\"Screenshot 2021-02-15 at 6 26 12 PM\"></p>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107950221-6e4ce500-6fbc-11eb-9900-91241aa73451.png\" alt=\"Screenshot 2021-02-15 at 6 34 00 PM\"></p>\n<p><img src=\"https://user-images.githubusercontent.com/40122399/107952170-693d6500-6fbf-11eb-8a8c-e93b54c0f693.png\" alt=\"Screenshot 2021-02-15 at 6 55 14 PM\"></p>\n<p>Thanks,<br>\nArindam</p>",
      "rawMarkdown": "Hello!!\n\nI am calculating the probability for each second, now for a clip of 60 sec, there will be 60 probabilities\n\n```\nmy 1 probability= probability of all 24 species for a 1 sec clip\n```\n![Screenshot 2021-02-15 at 6 26 12 PM](https://user-images.githubusercontent.com/40122399/107949494-5a54b380-6fbb-11eb-9027-c030673ccf86.png)\n\n\n![Screenshot 2021-02-15 at 6 34 00 PM](https://user-images.githubusercontent.com/40122399/107950221-6e4ce500-6fbc-11eb-9900-91241aa73451.png)\n\n![Screenshot 2021-02-15 at 6 55 14 PM](https://user-images.githubusercontent.com/40122399/107952170-693d6500-6fbf-11eb-8a8c-e93b54c0f693.png)\n\n\nThanks,\nArindam",
      "votes": null
    },
    {
      "id": "1201523",
      "postDate": "02/15/2021 13:16:41",
      "content": "<p>Max, therefore you can captures call detected in short intervals as well as those detected in long intervals.</p>",
      "rawMarkdown": "Max, therefore you can captures call detected in short intervals as well as those detected in long intervals.",
      "votes": null
    },
    {
      "id": "1201652",
      "postDate": "02/15/2021 14:46:10",
      "content": "<p>it is trial and error:</p>\n<ul>\n<li>max of mean (max over all models,  each model take means over 60 sec)</li>\n<li>mean of max</li>\n</ul>\n<p>in my experiments, mean of max always work better</p>",
      "rawMarkdown": "it is trial and error:\n- max of mean (max over all models,  each model take means over 60 sec)\n- mean of max\n\nin my experiments, mean of max always work better",
      "votes": null
    },
    {
      "id": "1201663",
      "postDate": "02/15/2021 14:53:48",
      "content": "<p>Thanks, it worked. Improved my score by 0.023 :)</p>",
      "rawMarkdown": "Thanks, it worked. Improved my score by 0.023 :)",
      "votes": null
    },
    {
      "id": "1201669",
      "postDate": "02/15/2021 14:58:27",
      "content": "<p>I appreciate your continued guidance, will surely try that :)</p>",
      "rawMarkdown": "I appreciate your continued guidance, will surely try that :)",
      "votes": null
    },
    {
      "id": "1207528",
      "postDate": "02/17/2021 22:34:06",
      "content": "<p>Update: tried max on max over two models, the score got improved by 0.021</p>",
      "rawMarkdown": "Update: tried max on max over two models, the score got improved by 0.021",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1201523,
      "author_name": "theoviel",
      "author_url": "",
      "post_date": "02/15/2021 13:16:41",
      "content": "<p>Max, therefore you can captures call detected in short intervals as well as those detected in long intervals.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1201663,
          "author_name": "jainarindam",
          "author_url": "",
          "post_date": "02/15/2021 14:53:48",
          "content": "<p>Thanks, it worked. Improved my score by 0.023 :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1201652,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "02/15/2021 14:46:10",
      "content": "<p>it is trial and error:</p>\n<ul>\n<li>max of mean (max over all models,  each model take means over 60 sec)</li>\n<li>mean of max</li>\n</ul>\n<p>in my experiments, mean of max always work better</p>",
      "votes": null,
      "replies": [
        {
          "id": 1201669,
          "author_name": "jainarindam",
          "author_url": "",
          "post_date": "02/15/2021 14:58:27",
          "content": "<p>I appreciate your continued guidance, will surely try that :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1207528,
          "author_name": "jainarindam",
          "author_url": "",
          "post_date": "02/17/2021 22:34:06",
          "content": "<p>Update: tried max on max over two models, the score got improved by 0.021</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1201472": "Hello!!\n\nI am calculating the probability for each second, now for a clip of 60 sec, there will be 60 probabilities\n\n```\nmy 1 probability= probability of all 24 species for a 1 sec clip\n```\n![Screenshot 2021-02-15 at 6 26 12 PM](https://user-images.githubusercontent.com/40122399/107949494-5a54b380-6fbb-11eb-9027-c030673ccf86.png)\n\n\n![Screenshot 2021-02-15 at 6 34 00 PM](https://user-images.githubusercontent.com/40122399/107950221-6e4ce500-6fbc-11eb-9900-91241aa73451.png)\n\n![Screenshot 2021-02-15 at 6 55 14 PM](https://user-images.githubusercontent.com/40122399/107952170-693d6500-6fbf-11eb-8a8c-e93b54c0f693.png)\n\n\nThanks,\nArindam",
    "1201523": "Max, therefore you can captures call detected in short intervals as well as those detected in long intervals.",
    "1201652": "it is trial and error:\n- max of mean (max over all models,  each model take means over 60 sec)\n- mean of max\n\nin my experiments, mean of max always work better",
    "1201663": "Thanks, it worked. Improved my score by 0.023 :)",
    "1201669": "I appreciate your continued guidance, will surely try that :)",
    "1207528": "Update: tried max on max over two models, the score got improved by 0.021"
  },
  "source": "meta"
}