{
  "id": 375820,
  "title": "Association Rule Mining is all we need?",
  "url": "/competitions/otto-recommender-system/discussion/375820",
  "author_name": "Anh Bui",
  "post_date": "2023-01-03T16:51:57.500000",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>After reading the idea from this article <a href=\"https://www.kaggle.com/code/cdeotte/candidate-rerank-model-lb-0-575\" target=\"_blank\">Baseline_Rerank</a>,  I think it is to explore 1:1 relationship between each pair of items. I'm thinking about how to build 2:1, 3:1, .. relationships between items. From there, I thought of Association Rules Mining to explore the characteristics of groups of items. Do you guys think this idea is feasible?</p>",
  "messages": [
    {
      "id": 2084668,
      "postDate": "2023-01-03T16:51:57.500Z",
      "content": "<p>After reading the idea from this article <a href=\"https://www.kaggle.com/code/cdeotte/candidate-rerank-model-lb-0-575\" target=\"_blank\">Baseline_Rerank</a>,  I think it is to explore 1:1 relationship between each pair of items. I'm thinking about how to build 2:1, 3:1, .. relationships between items. From there, I thought of Association Rules Mining to explore the characteristics of groups of items. Do you guys think this idea is feasible?</p>",
      "rawMarkdown": "\nAfter reading the idea from this article [Baseline_Rerank](https://www.kaggle.com/code/cdeotte/candidate-rerank-model-lb-0-575),  I think it is to explore 1:1 relationship between each pair of items. I'm thinking about how to build 2:1, 3:1, .. relationships between items. From there, I thought of Association Rules Mining to explore the characteristics of groups of items. Do you guys think this idea is feasible?",
      "votes": 4
    },
    {
      "id": 2085724,
      "postDate": "2023-01-04T10:51:28.613Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true,
      "replies": [
        {
          "id": 2085768,
          "postDate": "2023-01-04T11:44:36.293Z",
          "content": "<p>I can't use fpgrowth or apriori with such a large amount of data because the resources are not enough, is there any way to do it, at least for the 2:1 case?</p>",
          "rawMarkdown": "I can't use fpgrowth or apriori with such a large amount of data because the resources are not enough, is there any way to do it, at least for the 2:1 case?",
          "votes": 1,
          "replies": [
            {
              "id": 2085839,
              "postDate": "2023-01-04T12:35:43.873Z",
              "content": "<p>Hi Thiên đường tung tăng,</p>\n<p>I don't have much knowledge of possible algorithms for association that can work faster while also being less resource intensive but there are some optimizations than can be done on Apriori , you can learn more about it here :<br>\n<a href=\"https://www.degruyter.com/document/doi/10.1515/jisys-2020-0121/html?lang=en\" target=\"_blank\">https://www.degruyter.com/document/doi/10.1515/jisys-2020-0121/html?lang=en</a></p>\n<p>Hope it helps.</p>\n<p>Warm Regards,<br>\nBiswaroop</p>",
              "rawMarkdown": "Hi Thiên đường tung tăng,\n\nI don't have much knowledge of possible algorithms for association that can work faster while also being less resource intensive but there are some optimizations than can be done on Apriori , you can learn more about it here :\nhttps://www.degruyter.com/document/doi/10.1515/jisys-2020-0121/html?lang=en\n\nHope it helps.\n\nWarm Regards,\nBiswaroop",
              "votes": 1
            }
          ]
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2085724,
      "author_name": "",
      "author_url": "",
      "post_date": "2023-01-04T10:51:28.613000",
      "content": "",
      "votes": 1,
      "replies": [
        {
          "id": 2085768,
          "author_name": "Anh Bui",
          "author_url": "",
          "post_date": "2023-01-04T11:44:36.293000",
          "content": "<p>I can't use fpgrowth or apriori with such a large amount of data because the resources are not enough, is there any way to do it, at least for the 2:1 case?</p>",
          "votes": 1,
          "replies": [
            {
              "id": 2085839,
              "author_name": "Biswaroop Nath",
              "author_url": "",
              "post_date": "2023-01-04T12:35:43.873000",
              "content": "<p>Hi Thiên đường tung tăng,</p>\n<p>I don't have much knowledge of possible algorithms for association that can work faster while also being less resource intensive but there are some optimizations than can be done on Apriori , you can learn more about it here :<br>\n<a href=\"https://www.degruyter.com/document/doi/10.1515/jisys-2020-0121/html?lang=en\" target=\"_blank\">https://www.degruyter.com/document/doi/10.1515/jisys-2020-0121/html?lang=en</a></p>\n<p>Hope it helps.</p>\n<p>Warm Regards,<br>\nBiswaroop</p>",
              "votes": 1,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2084668": "\nAfter reading the idea from this article [Baseline_Rerank](https://www.kaggle.com/code/cdeotte/candidate-rerank-model-lb-0-575),  I think it is to explore 1:1 relationship between each pair of items. I'm thinking about how to build 2:1, 3:1, .. relationships between items. From there, I thought of Association Rules Mining to explore the characteristics of groups of items. Do you guys think this idea is feasible?",
    "2085724": ""
  }
}