{
  "id": 364942,
  "title": "Where is Customer and Products dataset for Customers Profiling and Products Profiling",
  "url": "/competitions/otto-recommender-system/discussion/364942",
  "author_name": "",
  "post_date": "2022-11-09T05:30:42.343062300Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi All,</p>\n<p>We are implementing any Customer Recommendation systems for e-com, we need to have customer profile by product and demographic ( there are more hybrid models)</p>\n<p>Do anybody have the customer and item data? Looks like we have only webshop transactions(Clicks,order, cart) dataset given?</p>\n<p>Without Customer profiling and Item profiling, how can you recommend a product to customer.</p>\n<p>aid - articles need some information like which brands, color, product category, price, on discount value <br>\nCustomer - age, area,pin code,gender, influencer </p>\n<p>With webshop tranactions, </p>\n<p>We can only get number of items/products clicked, ordered, browsered.  We even don't have $ of transaction, No return code or discount code information.</p>\n<p>Secondly, we only like to make single-stakeholder (Customer) or multi-stakeholder recommendation system (Customer, OTTO in this case)?</p>\n<p>Regards,<br>\nSandeep</p>",
  "messages": [
    {
      "id": "2022552",
      "postDate": "11/09/2022 05:30:42",
      "content": "<p>Hi All,</p>\n<p>We are implementing any Customer Recommendation systems for e-com, we need to have customer profile by product and demographic ( there are more hybrid models)</p>\n<p>Do anybody have the customer and item data? Looks like we have only webshop transactions(Clicks,order, cart) dataset given?</p>\n<p>Without Customer profiling and Item profiling, how can you recommend a product to customer.</p>\n<p>aid - articles need some information like which brands, color, product category, price, on discount value <br>\nCustomer - age, area,pin code,gender, influencer </p>\n<p>With webshop tranactions, </p>\n<p>We can only get number of items/products clicked, ordered, browsered.  We even don't have $ of transaction, No return code or discount code information.</p>\n<p>Secondly, we only like to make single-stakeholder (Customer) or multi-stakeholder recommendation system (Customer, OTTO in this case)?</p>\n<p>Regards,<br>\nSandeep</p>",
      "rawMarkdown": "Hi All,\n\nWe are implementing any Customer Recommendation systems for e-com, we need to have customer profile by product and demographic ( there are more hybrid models)\n\nDo anybody have the customer and item data? Looks like we have only webshop transactions(Clicks,order, cart) dataset given?\n\nWithout Customer profiling and Item profiling, how can you recommend a product to customer.\n\naid - articles need some information like which brands, color, product category, price, on discount value \nCustomer - age, area,pin code,gender, influencer \n\nWith webshop tranactions, \n\nWe can only get number of items/products clicked, ordered, browsered.  We even don't have $ of transaction, No return code or discount code information.\n\nSecondly, we only like to make single-stakeholder (Customer) or multi-stakeholder recommendation system (Customer, OTTO in this case)?\n\nRegards,\nSandeep",
      "votes": null
    },
    {
      "id": "2022641",
      "postDate": "11/09/2022 06:48:31",
      "content": "<p>There is a lot of information in the history of actions taken by a user 🙂 So far, we have only scratched the surface of what is possible on this data!</p>",
      "rawMarkdown": "There is a lot of information in the history of actions taken by a user 🙂 So far, we have only scratched the surface of what is possible on this data!",
      "votes": null
    },
    {
      "id": "2023103",
      "postDate": "11/09/2022 14:34:22",
      "content": "<p>Thanks Radek. The real world problems are different. We are only training Cross filter recommendation for Individual session. That's too with wrong assumption.</p>\n<p>If you are looking to buy a white addidas sneakers. Before Ordered(event type), do would have clicked <br>\nday 1<br>\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers <br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers<br>\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers<br>\naid:4 - adidas Originals »FORUM BONEGA MID« sneakers<br>\naid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker<br>\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers</p>\n<p>day2<br>\nCart: <br>\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers <br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers<br>\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers<br>\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers<br>\nOrdered:<br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129</p>\n<p>Day 5<br>\nAs per my understanding, recommendation system will predict as per your history:</p>\n<ol>\n<li>aid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129</li>\n<li>aid:1 - ADIDAS ORIGINALS »ZX 500« sneakers </li>\n<li>aid:3 - ADIDAS ORIGINALS»Superstar« sneakers</li>\n<li>aid:6 - Vans »SK8-Hi Platform 2.0« sneakers</li>\n<li>aid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker</li>\n<li>aid:4 - adidas Originals »FORUM BONEGA MID« sneakers</li>\n</ol>\n<p>If you have ordered the white sneaker of adidas, why recommendation system will like to sell you again \"sneaker\" from your own history? Why a user will buy a sneaker again?? Unless I haven't understood the problem statement.<br>\nSame goes with any high value item like bicycle, Sport equipment, Mulitmedia items  etc. </p>",
      "rawMarkdown": "Thanks Radek. The real world problems are different. We are only training Cross filter recommendation for Individual session. That's too with wrong assumption.\n\nIf you are looking to buy a white addidas sneakers. Before Ordered(event type), do would have clicked \nday 1\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers\naid:4 - adidas Originals »FORUM BONEGA MID« sneakers\naid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers\n\nday2\nCart: \naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers\nOrdered:\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129\n\nDay 5\nAs per my understanding, recommendation system will predict as per your history:\n1.  aid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129\n2. aid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \n3. aid:3 - ADIDAS ORIGINALS»Superstar« sneakers\n4. aid:6 - Vans »SK8-Hi Platform 2.0« sneakers\n5. aid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker\n6. aid:4 - adidas Originals »FORUM BONEGA MID« sneakers\n\nIf you have ordered the white sneaker of adidas, why recommendation system will like to sell you again \"sneaker\" from your own history? Why a user will buy a sneaker again?? Unless I haven't understood the problem statement.\nSame goes with any high value item like bicycle, Sport equipment, Mulitmedia items  etc.",
      "votes": null
    },
    {
      "id": "2023828",
      "postDate": "11/10/2022 04:20:38",
      "content": "<p>I agree with you that the data you have available will to a large extent dictate what you can do on a given problem 🙂 But I am quite fond of the formulation in this competition -- I feel the simplicity of data can lead to very creative solutions, pushing the boundary a little bit of what off-the-shelf approach could achieve if applied to this problem 🙂</p>\n<p>I am highly optimistic about how far we can get in this competition (the organizer didn't do anything silly like providing an unreasonable test set, tried some weird data obfuscation, etc), so I am quite excited for the competition and the degrees of freedom that this data allows 🙂 </p>\n<p>But I understand that some people might have appetite for solving a different recsys problem, and that is all fine. But I don't think it takes anything away from this competition and this dataset 🙂 I am already extremely grateful to the organizers for give us an opportunity to work on an interesting challenge such as this 🙂</p>",
      "rawMarkdown": "I agree with you that the data you have available will to a large extent dictate what you can do on a given problem 🙂 But I am quite fond of the formulation in this competition -- I feel the simplicity of data can lead to very creative solutions, pushing the boundary a little bit of what off-the-shelf approach could achieve if applied to this problem 🙂\n\nI am highly optimistic about how far we can get in this competition (the organizer didn't do anything silly like providing an unreasonable test set, tried some weird data obfuscation, etc), so I am quite excited for the competition and the degrees of freedom that this data allows 🙂 \n\nBut I understand that some people might have appetite for solving a different recsys problem, and that is all fine. But I don't think it takes anything away from this competition and this dataset 🙂 I am already extremely grateful to the organizers for give us an opportunity to work on an interesting challenge such as this 🙂",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2022641,
      "author_name": "radek1",
      "author_url": "",
      "post_date": "11/09/2022 06:48:31",
      "content": "<p>There is a lot of information in the history of actions taken by a user 🙂 So far, we have only scratched the surface of what is possible on this data!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2023103,
      "author_name": "x2y2z2xyz3",
      "author_url": "",
      "post_date": "11/09/2022 14:34:22",
      "content": "<p>Thanks Radek. The real world problems are different. We are only training Cross filter recommendation for Individual session. That's too with wrong assumption.</p>\n<p>If you are looking to buy a white addidas sneakers. Before Ordered(event type), do would have clicked <br>\nday 1<br>\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers <br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers<br>\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers<br>\naid:4 - adidas Originals »FORUM BONEGA MID« sneakers<br>\naid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker<br>\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers</p>\n<p>day2<br>\nCart: <br>\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers <br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers<br>\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers<br>\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers<br>\nOrdered:<br>\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129</p>\n<p>Day 5<br>\nAs per my understanding, recommendation system will predict as per your history:</p>\n<ol>\n<li>aid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129</li>\n<li>aid:1 - ADIDAS ORIGINALS »ZX 500« sneakers </li>\n<li>aid:3 - ADIDAS ORIGINALS»Superstar« sneakers</li>\n<li>aid:6 - Vans »SK8-Hi Platform 2.0« sneakers</li>\n<li>aid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker</li>\n<li>aid:4 - adidas Originals »FORUM BONEGA MID« sneakers</li>\n</ol>\n<p>If you have ordered the white sneaker of adidas, why recommendation system will like to sell you again \"sneaker\" from your own history? Why a user will buy a sneaker again?? Unless I haven't understood the problem statement.<br>\nSame goes with any high value item like bicycle, Sport equipment, Mulitmedia items  etc. </p>",
      "votes": null,
      "replies": [
        {
          "id": 2023828,
          "author_name": "radek1",
          "author_url": "",
          "post_date": "11/10/2022 04:20:38",
          "content": "<p>I agree with you that the data you have available will to a large extent dictate what you can do on a given problem 🙂 But I am quite fond of the formulation in this competition -- I feel the simplicity of data can lead to very creative solutions, pushing the boundary a little bit of what off-the-shelf approach could achieve if applied to this problem 🙂</p>\n<p>I am highly optimistic about how far we can get in this competition (the organizer didn't do anything silly like providing an unreasonable test set, tried some weird data obfuscation, etc), so I am quite excited for the competition and the degrees of freedom that this data allows 🙂 </p>\n<p>But I understand that some people might have appetite for solving a different recsys problem, and that is all fine. But I don't think it takes anything away from this competition and this dataset 🙂 I am already extremely grateful to the organizers for give us an opportunity to work on an interesting challenge such as this 🙂</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2022552": "Hi All,\n\nWe are implementing any Customer Recommendation systems for e-com, we need to have customer profile by product and demographic ( there are more hybrid models)\n\nDo anybody have the customer and item data? Looks like we have only webshop transactions(Clicks,order, cart) dataset given?\n\nWithout Customer profiling and Item profiling, how can you recommend a product to customer.\n\naid - articles need some information like which brands, color, product category, price, on discount value \nCustomer - age, area,pin code,gender, influencer \n\nWith webshop tranactions, \n\nWe can only get number of items/products clicked, ordered, browsered.  We even don't have $ of transaction, No return code or discount code information.\n\nSecondly, we only like to make single-stakeholder (Customer) or multi-stakeholder recommendation system (Customer, OTTO in this case)?\n\nRegards,\nSandeep",
    "2022641": "There is a lot of information in the history of actions taken by a user 🙂 So far, we have only scratched the surface of what is possible on this data!",
    "2023103": "Thanks Radek. The real world problems are different. We are only training Cross filter recommendation for Individual session. That's too with wrong assumption.\n\nIf you are looking to buy a white addidas sneakers. Before Ordered(event type), do would have clicked \nday 1\naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers\naid:4 - adidas Originals »FORUM BONEGA MID« sneakers\naid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers\n\nday2\nCart: \naid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers\naid:3 - ADIDAS ORIGINALS»Superstar« sneakers\naid:6 - Vans »SK8-Hi Platform 2.0« sneakers\nOrdered:\naid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129\n\nDay 5\nAs per my understanding, recommendation system will predict as per your history:\n1.  aid:2 - ADIDAS ORIGINALS »FORUM BOLD« sneakers - euro 129\n2. aid:1 - ADIDAS ORIGINALS »ZX 500« sneakers \n3. aid:3 - ADIDAS ORIGINALS»Superstar« sneakers\n4. aid:6 - Vans »SK8-Hi Platform 2.0« sneakers\n5. aid:5 - Nike Sportswear Wmns Court Royale 2 Mid Sneaker\n6. aid:4 - adidas Originals »FORUM BONEGA MID« sneakers\n\nIf you have ordered the white sneaker of adidas, why recommendation system will like to sell you again \"sneaker\" from your own history? Why a user will buy a sneaker again?? Unless I haven't understood the problem statement.\nSame goes with any high value item like bicycle, Sport equipment, Mulitmedia items  etc.",
    "2023828": "I agree with you that the data you have available will to a large extent dictate what you can do on a given problem 🙂 But I am quite fond of the formulation in this competition -- I feel the simplicity of data can lead to very creative solutions, pushing the boundary a little bit of what off-the-shelf approach could achieve if applied to this problem 🙂\n\nI am highly optimistic about how far we can get in this competition (the organizer didn't do anything silly like providing an unreasonable test set, tried some weird data obfuscation, etc), so I am quite excited for the competition and the degrees of freedom that this data allows 🙂 \n\nBut I understand that some people might have appetite for solving a different recsys problem, and that is all fine. But I don't think it takes anything away from this competition and this dataset 🙂 I am already extremely grateful to the organizers for give us an opportunity to work on an interesting challenge such as this 🙂"
  },
  "source": "meta"
}