{
  "id": 21928,
  "title": "Missing demand info",
  "url": "/competitions/grupo-bimbo-inventory-demand/discussion/21928",
  "author_name": "",
  "post_date": "2016-06-28T13:17:20.730Z",
  "votes": null,
  "comment_count": 4,
  "views": 723,
  "content": "<p>What are some of the strategies one can use for products that are in test but not in train (new products). How would one go by estimating demand for these products? </p>",
  "messages": [
    {
      "id": "125294",
      "postDate": "06/28/2016 13:17:20",
      "content": "<p>What are some of the strategies one can use for products that are in test but not in train (new products). How would one go by estimating demand for these products? </p>",
      "rawMarkdown": "What are some of the strategies one can use for products that are in test but not in train (new products). How would one go by estimating demand for these products?",
      "votes": null
    },
    {
      "id": "125297",
      "postDate": "06/28/2016 13:30:16",
      "content": "<p>Clustering Products using KMeans is a simple way of handling missing data</p>",
      "rawMarkdown": "Clustering Products using KMeans is a simple way of handling missing data",
      "votes": null
    },
    {
      "id": "125304",
      "postDate": "06/28/2016 14:05:53",
      "content": "<p>I'm not sure how effective standard method to impute missing data are in this case if there are products in test that we have no demand information for in training data. </p>\n\n<p>FR</p>",
      "rawMarkdown": "I'm not sure how effective standard method to impute missing data are in this case if there are products in test that we have no demand information for in training data. \r\n\r\nFR",
      "votes": null
    },
    {
      "id": "125317",
      "postDate": "06/28/2016 15:40:10",
      "content": "<p>Would you expect returns for similar types of products to be similar or dissimilar - perhaps you can use unsupervised learning so you don't need to know the demand to cluster - <a href=\"http://brandonrose.org/clustering\">link</a> </p>",
      "rawMarkdown": "Would you expect returns for similar types of products to be similar or dissimilar - perhaps you can use unsupervised learning so you don't need to know the demand to cluster - [link][1] \r\n\r\n\r\n  [1]: http://brandonrose.org/clustering",
      "votes": null
    },
    {
      "id": "125502",
      "postDate": "06/30/2016 02:34:43",
      "content": "<p>For the new products, maybe \n1. similarity can be learnt by the 'name' of products? For example, 'whatever old chocolate' might be similar with 'whatever new chocolate' in demand?</p>\n\n<ol start=\"2\">\n<li>'Depot- Client' association. Depot may have regular amounts of products sent to a certain customer</li>\n</ol>",
      "rawMarkdown": "For the new products, maybe \r\n1. similarity can be learnt by the 'name' of products? For example, 'whatever old chocolate' might be similar with 'whatever new chocolate' in demand?\r\n\r\n2. 'Depot- Client' association. Depot may have regular amounts of products sent to a certain customer",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 125297,
      "author_name": "scirpus",
      "author_url": "",
      "post_date": "06/28/2016 13:30:16",
      "content": "<p>Clustering Products using KMeans is a simple way of handling missing data</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 125304,
      "author_name": "tripletdad",
      "author_url": "",
      "post_date": "06/28/2016 14:05:53",
      "content": "<p>I'm not sure how effective standard method to impute missing data are in this case if there are products in test that we have no demand information for in training data. </p>\n\n<p>FR</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 125317,
      "author_name": "scirpus",
      "author_url": "",
      "post_date": "06/28/2016 15:40:10",
      "content": "<p>Would you expect returns for similar types of products to be similar or dissimilar - perhaps you can use unsupervised learning so you don't need to know the demand to cluster - <a href=\"http://brandonrose.org/clustering\">link</a> </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 125502,
      "author_name": "julesshen",
      "author_url": "",
      "post_date": "06/30/2016 02:34:43",
      "content": "<p>For the new products, maybe \n1. similarity can be learnt by the 'name' of products? For example, 'whatever old chocolate' might be similar with 'whatever new chocolate' in demand?</p>\n\n<ol start=\"2\">\n<li>'Depot- Client' association. Depot may have regular amounts of products sent to a certain customer</li>\n</ol>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "125294": "What are some of the strategies one can use for products that are in test but not in train (new products). How would one go by estimating demand for these products?",
    "125297": "Clustering Products using KMeans is a simple way of handling missing data",
    "125304": "I'm not sure how effective standard method to impute missing data are in this case if there are products in test that we have no demand information for in training data. \r\n\r\nFR",
    "125317": "Would you expect returns for similar types of products to be similar or dissimilar - perhaps you can use unsupervised learning so you don't need to know the demand to cluster - [link][1] \r\n\r\n\r\n  [1]: http://brandonrose.org/clustering",
    "125502": "For the new products, maybe \r\n1. similarity can be learnt by the 'name' of products? For example, 'whatever old chocolate' might be similar with 'whatever new chocolate' in demand?\r\n\r\n2. 'Depot- Client' association. Depot may have regular amounts of products sent to a certain customer"
  },
  "source": "meta"
}