{
  "id": 43083,
  "title": "Stock data",
  "url": "/competitions/favorita-grocery-sales-forecasting/discussion/43083",
  "author_name": "",
  "post_date": "2017-11-09T15:01:41.820238500Z",
  "votes": 6,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Currently, I have some doubts about the test set. <br>\nAt first, I supposed that all item/store combos in the test set were \"sellable\". Which means that they are all in stock. <br>\nThen I looked more precisely at some example and now I'm quite sure that some items are out of stock. <br>\nAnd without this information, it is not possible to decide if the unit_sales will be 0 or not.  :-/</p>\n\n<p>For instance: store_nbr = 4 and item_nbr = 1159728. <br>\n<img src=\"https://preview.ibb.co/dYoPdG/2017_11_09_15h41_30.png\" alt=\"unit_sales for 4/1159728\" title=\"\"> <br>\nAs we can see: when the product is in stock the sales are quite good (around 10). But there is no data after the end of 2015. <br>\nSo 2 possibles cases: either the product is out of stock, or the data is missing. <br>\nSo what am I suppose to forecast? 0 because I guess that the product is out of stock or something around 10 because I suspect a data issue (or around 10 because the product is once again in stock)?</p>\n\n<p>Am I missing something? </p>",
  "messages": [
    {
      "id": "241675",
      "postDate": "11/09/2017 15:01:41",
      "content": "<p>Currently, I have some doubts about the test set. <br>\nAt first, I supposed that all item/store combos in the test set were \"sellable\". Which means that they are all in stock. <br>\nThen I looked more precisely at some example and now I'm quite sure that some items are out of stock. <br>\nAnd without this information, it is not possible to decide if the unit_sales will be 0 or not.  :-/</p>\n\n<p>For instance: store_nbr = 4 and item_nbr = 1159728. <br>\n<img src=\"https://preview.ibb.co/dYoPdG/2017_11_09_15h41_30.png\" alt=\"unit_sales for 4/1159728\" title=\"\"> <br>\nAs we can see: when the product is in stock the sales are quite good (around 10). But there is no data after the end of 2015. <br>\nSo 2 possibles cases: either the product is out of stock, or the data is missing. <br>\nSo what am I suppose to forecast? 0 because I guess that the product is out of stock or something around 10 because I suspect a data issue (or around 10 because the product is once again in stock)?</p>\n\n<p>Am I missing something? </p>",
      "rawMarkdown": "Currently, I have some doubts about the test set.   \nAt first, I supposed that all item/store combos in the test set were \"sellable\". Which means that they are all in stock.  \nThen I looked more precisely at some example and now I'm quite sure that some items are out of stock.  \nAnd without this information, it is not possible to decide if the unit_sales will be 0 or not.  :-/\n\nFor instance: store_nbr = 4 and item_nbr = 1159728.  \n![unit_sales for 4/1159728][1]  \nAs we can see: when the product is in stock the sales are quite good (around 10). But there is no data after the end of 2015.  \nSo 2 possibles cases: either the product is out of stock, or the data is missing.  \nSo what am I suppose to forecast? 0 because I guess that the product is out of stock or something around 10 because I suspect a data issue (or around 10 because the product is once again in stock)?\n\nAm I missing something? \n\n  [1]: https://preview.ibb.co/dYoPdG/2017_11_09_15h41_30.png",
      "votes": null
    },
    {
      "id": "241717",
      "postDate": "11/09/2017 16:57:49",
      "content": "<p>Good spot. Many reasons can explain this phenomenon:</p>\n\n<ul>\n<li><p>The item is out of stock (as you said)</p></li>\n<li><p>The shell of that item is in reparation so no item is shown. When the shell is ready, the item is for sale again.</p></li>\n<li><p>The item is end of life (the producer replaces it by a new product)</p></li>\n<li><p>The product is not relatively profitable so the stock manager decides to stop selling it</p></li>\n</ul>\n\n<p>Of course in this case the prediction should be indeed zero.</p>",
      "rawMarkdown": "Good spot. Many reasons can explain this phenomenon:\n\n- The item is out of stock (as you said)\n\n- The shell of that item is in reparation so no item is shown. When the shell is ready, the item is for sale again.\n\n- The item is end of life (the producer replaces it by a new product)\n\n- The product is not relatively profitable so the stock manager decides to stop selling it\n\n\nOf course in this case the prediction should be indeed zero.",
      "votes": null
    },
    {
      "id": "241763",
      "postDate": "11/09/2017 18:57:57",
      "content": "<p>One data cleaning step I took was to only look at store/item combinations found in both the training set and the test set. However, you make a good point about stock levels.  I have no idea if an item is in stock.</p>",
      "rawMarkdown": "One data cleaning step I took was to only look at store/item combinations found in both the training set and the test set. However, you make a good point about stock levels.  I have no idea if an item is in stock.",
      "votes": null
    },
    {
      "id": "242633",
      "postDate": "11/12/2017 07:43:03",
      "content": "<p>I noticed this too. While I'm not surprised we have no information on whether an item is in/out-of-stock or on/off-the-shelf, it is quite annoying. However, I have bigger fish to fry at the moment --like creating an actual decent baseline. :P</p>",
      "rawMarkdown": "I noticed this too. While I'm not surprised we have no information on whether an item is in/out-of-stock or on/off-the-shelf, it is quite annoying. However, I have bigger fish to fry at the moment --like creating an actual decent baseline. :P",
      "votes": null
    },
    {
      "id": "243094",
      "postDate": "11/13/2017 10:55:10",
      "content": "<p>:-) <br>\nYou know, your baseline is hugely impacted by in/out-of-stock and on/off-the-shelf. <br>\nDoes a drop in the sales is related to the oil price, the holiday or the fact that the product is off-the-shelf at 4 pm because of low stock? :-/</p>",
      "rawMarkdown": ":-)  \nYou know, your baseline is hugely impacted by in/out-of-stock and on/off-the-shelf.  \nDoes a drop in the sales is related to the oil price, the holiday or the fact that the product is off-the-shelf at 4 pm because of low stock? :-/",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 241717,
      "author_name": "tarobxl",
      "author_url": "",
      "post_date": "11/09/2017 16:57:49",
      "content": "<p>Good spot. Many reasons can explain this phenomenon:</p>\n\n<ul>\n<li><p>The item is out of stock (as you said)</p></li>\n<li><p>The shell of that item is in reparation so no item is shown. When the shell is ready, the item is for sale again.</p></li>\n<li><p>The item is end of life (the producer replaces it by a new product)</p></li>\n<li><p>The product is not relatively profitable so the stock manager decides to stop selling it</p></li>\n</ul>\n\n<p>Of course in this case the prediction should be indeed zero.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 241763,
      "author_name": "bdub76",
      "author_url": "",
      "post_date": "11/09/2017 18:57:57",
      "content": "<p>One data cleaning step I took was to only look at store/item combinations found in both the training set and the test set. However, you make a good point about stock levels.  I have no idea if an item is in stock.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 242633,
      "author_name": "puremath86",
      "author_url": "",
      "post_date": "11/12/2017 07:43:03",
      "content": "<p>I noticed this too. While I'm not surprised we have no information on whether an item is in/out-of-stock or on/off-the-shelf, it is quite annoying. However, I have bigger fish to fry at the moment --like creating an actual decent baseline. :P</p>",
      "votes": null,
      "replies": [
        {
          "id": 243094,
          "author_name": "ericdfr",
          "author_url": "",
          "post_date": "11/13/2017 10:55:10",
          "content": "<p>:-) <br>\nYou know, your baseline is hugely impacted by in/out-of-stock and on/off-the-shelf. <br>\nDoes a drop in the sales is related to the oil price, the holiday or the fact that the product is off-the-shelf at 4 pm because of low stock? :-/</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "241675": "Currently, I have some doubts about the test set.   \nAt first, I supposed that all item/store combos in the test set were \"sellable\". Which means that they are all in stock.  \nThen I looked more precisely at some example and now I'm quite sure that some items are out of stock.  \nAnd without this information, it is not possible to decide if the unit_sales will be 0 or not.  :-/\n\nFor instance: store_nbr = 4 and item_nbr = 1159728.  \n![unit_sales for 4/1159728][1]  \nAs we can see: when the product is in stock the sales are quite good (around 10). But there is no data after the end of 2015.  \nSo 2 possibles cases: either the product is out of stock, or the data is missing.  \nSo what am I suppose to forecast? 0 because I guess that the product is out of stock or something around 10 because I suspect a data issue (or around 10 because the product is once again in stock)?\n\nAm I missing something? \n\n  [1]: https://preview.ibb.co/dYoPdG/2017_11_09_15h41_30.png",
    "241717": "Good spot. Many reasons can explain this phenomenon:\n\n- The item is out of stock (as you said)\n\n- The shell of that item is in reparation so no item is shown. When the shell is ready, the item is for sale again.\n\n- The item is end of life (the producer replaces it by a new product)\n\n- The product is not relatively profitable so the stock manager decides to stop selling it\n\n\nOf course in this case the prediction should be indeed zero.",
    "241763": "One data cleaning step I took was to only look at store/item combinations found in both the training set and the test set. However, you make a good point about stock levels.  I have no idea if an item is in stock.",
    "242633": "I noticed this too. While I'm not surprised we have no information on whether an item is in/out-of-stock or on/off-the-shelf, it is quite annoying. However, I have bigger fish to fry at the moment --like creating an actual decent baseline. :P",
    "243094": ":-)  \nYou know, your baseline is hugely impacted by in/out-of-stock and on/off-the-shelf.  \nDoes a drop in the sales is related to the oil price, the holiday or the fact that the product is off-the-shelf at 4 pm because of low stock? :-/"
  },
  "source": "meta"
}