{
  "id": 56693,
  "title": "What does the field 'item_seq_number' mean ?",
  "url": "/competitions/avito-demand-prediction/discussion/56693",
  "author_name": "",
  "post_date": "2018-05-13T15:56:20.930214800Z",
  "votes": 7,
  "comment_count": 8,
  "views": 0,
  "content": "<p>As the title says, I failed to understand meaning of the attribute item_seq_number - Ad sequential number for user. Can somebody please explain what it means in this context</p>",
  "messages": [
    {
      "id": "328191",
      "postDate": "05/13/2018 15:56:20",
      "content": "<p>As the title says, I failed to understand meaning of the attribute item_seq_number - Ad sequential number for user. Can somebody please explain what it means in this context</p>",
      "rawMarkdown": "As the title says, I failed to understand meaning of the attribute item_seq_number - Ad sequential number for user. Can somebody please explain what it means in this context",
      "votes": null
    },
    {
      "id": "328197",
      "postDate": "05/13/2018 16:16:35",
      "content": "<p>I fell it has something to do with sequential advertising.</p>\n\n<pre><code>sequential advertising guarantees a person will see Ad 1, then Ad 2, then Ad 3, and so on, even if they’re switching between many devices. \n</code></pre>\n\n<p><a href=\"https://blog.adstage.io/2017/05/10/facebook-sequential-advertising/\">source and to read more.</a></p>",
      "rawMarkdown": "I fell it has something to do with sequential advertising.\n\n    sequential advertising guarantees a person will see Ad 1, then Ad 2, then Ad 3, and so on, even if they’re switching between many devices. \n[source and to read more.][1]\n\n\n  [1]: https://blog.adstage.io/2017/05/10/facebook-sequential-advertising/",
      "votes": null
    },
    {
      "id": "328198",
      "postDate": "05/13/2018 16:21:15",
      "content": "<p>And since according to <a href=\"https://www.kaggle.com/c/avito-demand-prediction/discussion/56531\">this discussion</a></p>\n\n<pre><code> Avito computes deal probability based on a combination of how many people view the ad and how many people click the button. \n</code></pre>\n\n<p>That's why ads with small seq_num have higher deal probability, <a href=\"https://www.kaggle.com/tapioca/item-seq-number-vs-deal-probability\">see this kernel</a></p>",
      "rawMarkdown": "And since according to [this discussion][1]\n\n     Avito computes deal probability based on a combination of how many people view the ad and how many people click the button. \nThat's why ads with small seq_num have higher deal probability, [see this kernel][2]\n\n\n  [1]: https://www.kaggle.com/c/avito-demand-prediction/discussion/56531\n  [2]: https://www.kaggle.com/tapioca/item-seq-number-vs-deal-probability",
      "votes": null
    },
    {
      "id": "328221",
      "postDate": "05/13/2018 17:50:07",
      "content": "<p>Thank you for the explanation ! </p>",
      "rawMarkdown": "Thank you for the explanation !",
      "votes": null
    },
    {
      "id": "328534",
      "postDate": "05/14/2018 15:05:00",
      "content": "<p>Thanx, it was on my for a while.</p>",
      "rawMarkdown": "Thanx, it was on my for a while.",
      "votes": null
    },
    {
      "id": "329639",
      "postDate": "05/16/2018 21:11:34",
      "content": "<p>I strongly suspect that it is a global cumulative count of the number of items/ads that a user has posted on Avito. So item_seq_number 1 would mean that's the first time a user has posted to the site, 10000 would mean that the user has posted 9999 items already, etc. </p>",
      "rawMarkdown": "I strongly suspect that it is a global cumulative count of the number of items/ads that a user has posted on Avito. So item_seq_number 1 would mean that's the first time a user has posted to the site, 10000 would mean that the user has posted 9999 items already, etc.",
      "votes": null
    },
    {
      "id": "329640",
      "postDate": "05/16/2018 21:13:27",
      "content": "<p>I suspect this too. We know that <code>item_seq_number</code> is time ordered and higher on average for <code>user_type</code> with more expected volume.</p>",
      "rawMarkdown": "I suspect this too. We know that `item_seq_number` is time ordered and higher on average for `user_type` with more expected volume.",
      "votes": null
    },
    {
      "id": "329978",
      "postDate": "05/17/2018 18:23:32",
      "content": "<p>I agree, it's pretty simple to validate this : I did a grouping by user and counted the number of Ads, I had a number often equals to item_seq_number. In case where the user have more than 1000 they might be difference, probably because not all data are in train.csv. But this seems to confirm that it's a simple count of user ads.</p>",
      "rawMarkdown": "I agree, it's pretty simple to validate this : I did a grouping by user and counted the number of Ads, I had a number often equals to item_seq_number. In case where the user have more than 1000 they might be difference, probably because not all data are in train.csv. But this seems to confirm that it's a simple count of user ads.",
      "votes": null
    },
    {
      "id": "334930",
      "postDate": "05/28/2018 18:28:06",
      "content": "<p>Although looking at train_active, I was able to find user_ids with duplicate item_seq_number values</p>",
      "rawMarkdown": "Although looking at train_active, I was able to find user_ids with duplicate item_seq_number values",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 328197,
      "author_name": "sohaibomar",
      "author_url": "",
      "post_date": "05/13/2018 16:16:35",
      "content": "<p>I fell it has something to do with sequential advertising.</p>\n\n<pre><code>sequential advertising guarantees a person will see Ad 1, then Ad 2, then Ad 3, and so on, even if they’re switching between many devices. \n</code></pre>\n\n<p><a href=\"https://blog.adstage.io/2017/05/10/facebook-sequential-advertising/\">source and to read more.</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 328198,
          "author_name": "sohaibomar",
          "author_url": "",
          "post_date": "05/13/2018 16:21:15",
          "content": "<p>And since according to <a href=\"https://www.kaggle.com/c/avito-demand-prediction/discussion/56531\">this discussion</a></p>\n\n<pre><code> Avito computes deal probability based on a combination of how many people view the ad and how many people click the button. \n</code></pre>\n\n<p>That's why ads with small seq_num have higher deal probability, <a href=\"https://www.kaggle.com/tapioca/item-seq-number-vs-deal-probability\">see this kernel</a></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 328221,
          "author_name": "pasumarthysrikanth",
          "author_url": "",
          "post_date": "05/13/2018 17:50:07",
          "content": "<p>Thank you for the explanation ! </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 328534,
          "author_name": "kainsama",
          "author_url": "",
          "post_date": "05/14/2018 15:05:00",
          "content": "<p>Thanx, it was on my for a while.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 329639,
      "author_name": "aquatic",
      "author_url": "",
      "post_date": "05/16/2018 21:11:34",
      "content": "<p>I strongly suspect that it is a global cumulative count of the number of items/ads that a user has posted on Avito. So item_seq_number 1 would mean that's the first time a user has posted to the site, 10000 would mean that the user has posted 9999 items already, etc. </p>",
      "votes": null,
      "replies": [
        {
          "id": 329640,
          "author_name": "peterhurford",
          "author_url": "",
          "post_date": "05/16/2018 21:13:27",
          "content": "<p>I suspect this too. We know that <code>item_seq_number</code> is time ordered and higher on average for <code>user_type</code> with more expected volume.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 329978,
          "author_name": "areveillon",
          "author_url": "",
          "post_date": "05/17/2018 18:23:32",
          "content": "<p>I agree, it's pretty simple to validate this : I did a grouping by user and counted the number of Ads, I had a number often equals to item_seq_number. In case where the user have more than 1000 they might be difference, probably because not all data are in train.csv. But this seems to confirm that it's a simple count of user ads.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 334930,
          "author_name": "abhimanyud",
          "author_url": "",
          "post_date": "05/28/2018 18:28:06",
          "content": "<p>Although looking at train_active, I was able to find user_ids with duplicate item_seq_number values</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "328191": "As the title says, I failed to understand meaning of the attribute item_seq_number - Ad sequential number for user. Can somebody please explain what it means in this context",
    "328197": "I fell it has something to do with sequential advertising.\n\n    sequential advertising guarantees a person will see Ad 1, then Ad 2, then Ad 3, and so on, even if they’re switching between many devices. \n[source and to read more.][1]\n\n\n  [1]: https://blog.adstage.io/2017/05/10/facebook-sequential-advertising/",
    "328198": "And since according to [this discussion][1]\n\n     Avito computes deal probability based on a combination of how many people view the ad and how many people click the button. \nThat's why ads with small seq_num have higher deal probability, [see this kernel][2]\n\n\n  [1]: https://www.kaggle.com/c/avito-demand-prediction/discussion/56531\n  [2]: https://www.kaggle.com/tapioca/item-seq-number-vs-deal-probability",
    "328221": "Thank you for the explanation !",
    "328534": "Thanx, it was on my for a while.",
    "329639": "I strongly suspect that it is a global cumulative count of the number of items/ads that a user has posted on Avito. So item_seq_number 1 would mean that's the first time a user has posted to the site, 10000 would mean that the user has posted 9999 items already, etc.",
    "329640": "I suspect this too. We know that `item_seq_number` is time ordered and higher on average for `user_type` with more expected volume.",
    "329978": "I agree, it's pretty simple to validate this : I did a grouping by user and counted the number of Ads, I had a number often equals to item_seq_number. In case where the user have more than 1000 they might be difference, probably because not all data are in train.csv. But this seems to confirm that it's a simple count of user ads.",
    "334930": "Although looking at train_active, I was able to find user_ids with duplicate item_seq_number values"
  },
  "source": "meta"
}