{
  "id": 56537,
  "title": "Time ads was active - periods_train.csv",
  "url": "/competitions/avito-demand-prediction/discussion/56537",
  "author_name": "",
  "post_date": "2018-05-10T20:21:45.735431200Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hello,</p>\n\n<p>I was trying to compute the time the ad was active. Unfortunately we only have the deal_probability for train.csv but periods_train.csv is only for item_id in train_active.csv...</p>\n\n<p>Would it be possible to have this periods_train.csv for the train.csv items ?</p>\n\n<p>Thanks,</p>",
  "messages": [
    {
      "id": "327101",
      "postDate": "05/10/2018 20:21:45",
      "content": "<p>Hello,</p>\n\n<p>I was trying to compute the time the ad was active. Unfortunately we only have the deal_probability for train.csv but periods_train.csv is only for item_id in train_active.csv...</p>\n\n<p>Would it be possible to have this periods_train.csv for the train.csv items ?</p>\n\n<p>Thanks,</p>",
      "rawMarkdown": "Hello,\n\nI was trying to compute the time the ad was active. Unfortunately we only have the deal_probability for train.csv but periods_train.csv is only for item_id in train_active.csv...\n\nWould it be possible to have this periods_train.csv for the train.csv items ?\n\nThanks,",
      "votes": null
    },
    {
      "id": "327121",
      "postDate": "05/10/2018 20:56:23",
      "content": "<p>I have a feeling that it's not very likely - risk of leakage is too high from the organizers' pov.</p>",
      "rawMarkdown": "I have a feeling that it's not very likely - risk of leakage is too high from the organizers' pov.",
      "votes": null
    },
    {
      "id": "327125",
      "postDate": "05/10/2018 21:15:48",
      "content": "<p>I guess only price might be use full from train/test_active.csv. We might not learn the time ad was active from periods_train.csv for train set.</p>",
      "rawMarkdown": "I guess only price might be use full from train/test_active.csv. We might not learn the time ad was active from periods_train.csv for train set.",
      "votes": null
    },
    {
      "id": "327144",
      "postDate": "05/10/2018 22:38:22",
      "content": "<p>I haven't done this yet but I wonder if there are some common users between train and train_active. If so, we could get mean/median ad length for some users. It's on my list to try so if I do and it has any impact I'll report back.</p>",
      "rawMarkdown": "I haven't done this yet but I wonder if there are some common users between train and train_active. If so, we could get mean/median ad length for some users. It's on my list to try so if I do and it has any impact I'll report back.",
      "votes": null
    },
    {
      "id": "327422",
      "postDate": "05/11/2018 14:15:29",
      "content": "<p>Thanks for your answers... I guess if they provided the information , there might be some way to get some information from it, maybe by users As <a href=\"/michaelpawlus\">@michaelpawlus</a> was saying... I'll try that ! Thanks.</p>",
      "rawMarkdown": "Thanks for your answers... I guess if they provided the information , there might be some way to get some information from it, maybe by users As @michaelpawlus was saying... I'll try that ! Thanks.",
      "votes": null
    },
    {
      "id": "343822",
      "postDate": "06/16/2018 08:09:53",
      "content": "<p>You might want to check this out:\n<a href=\"https://www.kaggle.com/bminixhofer/aggregated-features-lightgbm\">https://www.kaggle.com/bminixhofer/aggregated-features-lightgbm</a></p>",
      "rawMarkdown": "You might want to check this out:\nhttps://www.kaggle.com/bminixhofer/aggregated-features-lightgbm",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 327121,
      "author_name": "konradb",
      "author_url": "",
      "post_date": "05/10/2018 20:56:23",
      "content": "<p>I have a feeling that it's not very likely - risk of leakage is too high from the organizers' pov.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 327125,
      "author_name": "sohaibomar",
      "author_url": "",
      "post_date": "05/10/2018 21:15:48",
      "content": "<p>I guess only price might be use full from train/test_active.csv. We might not learn the time ad was active from periods_train.csv for train set.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 327144,
      "author_name": "michaelpawlus",
      "author_url": "",
      "post_date": "05/10/2018 22:38:22",
      "content": "<p>I haven't done this yet but I wonder if there are some common users between train and train_active. If so, we could get mean/median ad length for some users. It's on my list to try so if I do and it has any impact I'll report back.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 327422,
      "author_name": "areveillon",
      "author_url": "",
      "post_date": "05/11/2018 14:15:29",
      "content": "<p>Thanks for your answers... I guess if they provided the information , there might be some way to get some information from it, maybe by users As <a href=\"/michaelpawlus\">@michaelpawlus</a> was saying... I'll try that ! Thanks.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 343822,
      "author_name": "iandzindo",
      "author_url": "",
      "post_date": "06/16/2018 08:09:53",
      "content": "<p>You might want to check this out:\n<a href=\"https://www.kaggle.com/bminixhofer/aggregated-features-lightgbm\">https://www.kaggle.com/bminixhofer/aggregated-features-lightgbm</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "327101": "Hello,\n\nI was trying to compute the time the ad was active. Unfortunately we only have the deal_probability for train.csv but periods_train.csv is only for item_id in train_active.csv...\n\nWould it be possible to have this periods_train.csv for the train.csv items ?\n\nThanks,",
    "327121": "I have a feeling that it's not very likely - risk of leakage is too high from the organizers' pov.",
    "327125": "I guess only price might be use full from train/test_active.csv. We might not learn the time ad was active from periods_train.csv for train set.",
    "327144": "I haven't done this yet but I wonder if there are some common users between train and train_active. If so, we could get mean/median ad length for some users. It's on my list to try so if I do and it has any impact I'll report back.",
    "327422": "Thanks for your answers... I guess if they provided the information , there might be some way to get some information from it, maybe by users As @michaelpawlus was saying... I'll try that ! Thanks.",
    "343822": "You might want to check this out:\nhttps://www.kaggle.com/bminixhofer/aggregated-features-lightgbm"
  },
  "source": "meta"
}