{
  "id": 204808,
  "title": "Conclusions",
  "url": "/competitions/predict-volcanic-eruptions-ingv-oe/discussion/204808",
  "author_name": "",
  "post_date": "2020-12-16T23:42:12.778744600Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hello!</p>\n<p>I'm new at this competition and I'm trying to understand what we are meant to do and to conclude after dealing with our data. Once we have our results, would we be able to predict when a volcano (or at least this particular volcano) is it going to erupt? I mean, I understand that for each set of data, we are going to obtain a time to eruption based on a perturbation of the signal of our sensors. I'm trying to see the problem in a realistic way, I don't know if you could help me. Would it be possible to extrapolate our results to another volcanoes? Thanks in advance!</p>",
  "messages": [
    {
      "id": "1116179",
      "postDate": "12/16/2020 23:42:12",
      "content": "<p>Hello!</p>\n<p>I'm new at this competition and I'm trying to understand what we are meant to do and to conclude after dealing with our data. Once we have our results, would we be able to predict when a volcano (or at least this particular volcano) is it going to erupt? I mean, I understand that for each set of data, we are going to obtain a time to eruption based on a perturbation of the signal of our sensors. I'm trying to see the problem in a realistic way, I don't know if you could help me. Would it be possible to extrapolate our results to another volcanoes? Thanks in advance!</p>",
      "rawMarkdown": "Hello!\n\nI'm new at this competition and I'm trying to understand what we are meant to do and to conclude after dealing with our data. Once we have our results, would we be able to predict when a volcano (or at least this particular volcano) is it going to erupt? I mean, I understand that for each set of data, we are going to obtain a time to eruption based on a perturbation of the signal of our sensors. I'm trying to see the problem in a realistic way, I don't know if you could help me. Would it be possible to extrapolate our results to another volcanoes? Thanks in advance!",
      "votes": null
    },
    {
      "id": "1117279",
      "postDate": "12/17/2020 21:47:02",
      "content": "<p>Hello <a href=\"https://www.kaggle.com/gatoschrodinger\" target=\"_blank\">@gatoschrodinger</a> </p>\n<p>Yes, the idea is to try to predict the time to the next eruption from the 10 minute sample of the sensors.</p>\n<p>Although I think you could manage to have an approach to try to predict the eruption time with real data, since it is real signals, even with a lot of missing data.</p>\n<p>In addition, I think that it would not be possible to predict eruptions of other volcanoes, nor that a prediction could be generalized from the data of this volcano. The main problem is the number of observations.</p>\n<p>Many of us have problems with overfitting particularly because of the number of samples and the time of each sample.</p>\n<p>I think this would be a first step to continue developing models with much more data and to achieve generalization.</p>",
      "rawMarkdown": "Hello @gatoschrodinger \n\nYes, the idea is to try to predict the time to the next eruption from the 10 minute sample of the sensors.\n\nAlthough I think you could manage to have an approach to try to predict the eruption time with real data, since it is real signals, even with a lot of missing data.\n\nIn addition, I think that it would not be possible to predict eruptions of other volcanoes, nor that a prediction could be generalized from the data of this volcano. The main problem is the number of observations.\n\nMany of us have problems with overfitting particularly because of the number of samples and the time of each sample.\n\nI think this would be a first step to continue developing models with much more data and to achieve generalization.",
      "votes": null
    },
    {
      "id": "1118750",
      "postDate": "12/19/2020 11:10:53",
      "content": "<p>Thank you so much!<br>\nI was just trying to understand exactly what we are meant to do, since I don't understand exactly what conclusions can we reach once data is analized. </p>",
      "rawMarkdown": "Thank you so much!\nI was just trying to understand exactly what we are meant to do, since I don't understand exactly what conclusions can we reach once data is analized.",
      "votes": null
    },
    {
      "id": "1118990",
      "postDate": "12/19/2020 15:56:40",
      "content": "<p>Desareca great !</p>",
      "rawMarkdown": "Desareca great !",
      "votes": null
    },
    {
      "id": "1119204",
      "postDate": "12/19/2020 20:08:53",
      "content": "<p>thanks!!! 👍</p>",
      "rawMarkdown": "thanks!!! 👍",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1117279,
      "author_name": "desareca",
      "author_url": "",
      "post_date": "12/17/2020 21:47:02",
      "content": "<p>Hello <a href=\"https://www.kaggle.com/gatoschrodinger\" target=\"_blank\">@gatoschrodinger</a> </p>\n<p>Yes, the idea is to try to predict the time to the next eruption from the 10 minute sample of the sensors.</p>\n<p>Although I think you could manage to have an approach to try to predict the eruption time with real data, since it is real signals, even with a lot of missing data.</p>\n<p>In addition, I think that it would not be possible to predict eruptions of other volcanoes, nor that a prediction could be generalized from the data of this volcano. The main problem is the number of observations.</p>\n<p>Many of us have problems with overfitting particularly because of the number of samples and the time of each sample.</p>\n<p>I think this would be a first step to continue developing models with much more data and to achieve generalization.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1118750,
          "author_name": "gatoschrodinger",
          "author_url": "",
          "post_date": "12/19/2020 11:10:53",
          "content": "<p>Thank you so much!<br>\nI was just trying to understand exactly what we are meant to do, since I don't understand exactly what conclusions can we reach once data is analized. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1118990,
          "author_name": "",
          "author_url": "",
          "post_date": "12/19/2020 15:56:40",
          "content": "<p>Desareca great !</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1119204,
          "author_name": "desareca",
          "author_url": "",
          "post_date": "12/19/2020 20:08:53",
          "content": "<p>thanks!!! 👍</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1116179": "Hello!\n\nI'm new at this competition and I'm trying to understand what we are meant to do and to conclude after dealing with our data. Once we have our results, would we be able to predict when a volcano (or at least this particular volcano) is it going to erupt? I mean, I understand that for each set of data, we are going to obtain a time to eruption based on a perturbation of the signal of our sensors. I'm trying to see the problem in a realistic way, I don't know if you could help me. Would it be possible to extrapolate our results to another volcanoes? Thanks in advance!",
    "1117279": "Hello @gatoschrodinger \n\nYes, the idea is to try to predict the time to the next eruption from the 10 minute sample of the sensors.\n\nAlthough I think you could manage to have an approach to try to predict the eruption time with real data, since it is real signals, even with a lot of missing data.\n\nIn addition, I think that it would not be possible to predict eruptions of other volcanoes, nor that a prediction could be generalized from the data of this volcano. The main problem is the number of observations.\n\nMany of us have problems with overfitting particularly because of the number of samples and the time of each sample.\n\nI think this would be a first step to continue developing models with much more data and to achieve generalization.",
    "1118750": "Thank you so much!\nI was just trying to understand exactly what we are meant to do, since I don't understand exactly what conclusions can we reach once data is analized.",
    "1118990": "Desareca great !",
    "1119204": "thanks!!! 👍"
  },
  "source": "meta"
}