{
  "id": 7680,
  "title": "Related useful models/modellings",
  "url": "/competitions/acquire-valued-shoppers-challenge/discussion/7680",
  "author_name": "",
  "post_date": "2014-04-13T10:23:25.883Z",
  "votes": 18,
  "comment_count": 3,
  "views": 4013,
  "content": "<p>Predicting whether customer becomes repeat buyer is &quot;sort of&quot; sub-task in estimation of&nbsp;customer's lifetime value. Best models in that field (to my understanding) are based on Bayesian models which model timing of first purchase, 2, 3, ... , amount in first purchase , 2, 3, ..., customer churning, etc etc.</p>\n<p>Remark that in this competition one does not need to do full lifetime value modelling (which would be, perhaps, a bit &quot;overkill&quot;).</p>\n<p>Anyways it might be worth taking&nbsp;a look at the following things:</p>\n<ul>\n<li><span style=\"line-height: 1.4\"><em>&quot;Modeling Customer Lifetime Value&quot;</em> http://www.anderson.ucla.edu/faculty/dominique.hanssens/content/JSR2006.pdf</span></li>\n<li><em>&quot;Measuring Customer Life Time Value: Models and Analysis&quot;</em> http://www.insead.edu/facultyresearch/research/doc.cfm?did=51835</li>\n<li><em>Custora's CLV related material:</em><br>https://www.custora.com/university/for-marketers/clv/advanced/bayesian-inference<br>and<br>https://www.custora.com/university/for-marketers/clv/advanced/pareto-nbd</li>\n<li>&quot;A modified Pareto/NBD approach for predicting customer lifetime value&quot;<br>http://www.essec.edu/faculty/showDeclFileRes.do?declId=8555&amp;key=Publication-Content</li>\n<li>&quot;Improved Pareto/NBD Model and Its Applications in Customer Segmentation based on PErsonal Information Combination&quot; http://www.sersc.org/journals/IJDTA/vol6_no5/16.pdf</li>\n</ul>\n<p>In above NBD is negative binomial distribution which was used in early phases of computer modelling . See wiki page&nbsp;http://en.wikipedia.org/wiki/Negative_binomial_distribution<br>Remark that NBD modelling approach has flaws but it is better than nothing :)</p>",
  "messages": [
    {
      "id": "41840",
      "postDate": "04/13/2014 10:23:25",
      "content": "<p>Predicting whether customer becomes repeat buyer is &quot;sort of&quot; sub-task in estimation of&nbsp;customer's lifetime value. Best models in that field (to my understanding) are based on Bayesian models which model timing of first purchase, 2, 3, ... , amount in first purchase , 2, 3, ..., customer churning, etc etc.</p>\n<p>Remark that in this competition one does not need to do full lifetime value modelling (which would be, perhaps, a bit &quot;overkill&quot;).</p>\n<p>Anyways it might be worth taking&nbsp;a look at the following things:</p>\n<ul>\n<li><span style=\"line-height: 1.4\"><em>&quot;Modeling Customer Lifetime Value&quot;</em> http://www.anderson.ucla.edu/faculty/dominique.hanssens/content/JSR2006.pdf</span></li>\n<li><em>&quot;Measuring Customer Life Time Value: Models and Analysis&quot;</em> http://www.insead.edu/facultyresearch/research/doc.cfm?did=51835</li>\n<li><em>Custora's CLV related material:</em><br>https://www.custora.com/university/for-marketers/clv/advanced/bayesian-inference<br>and<br>https://www.custora.com/university/for-marketers/clv/advanced/pareto-nbd</li>\n<li>&quot;A modified Pareto/NBD approach for predicting customer lifetime value&quot;<br>http://www.essec.edu/faculty/showDeclFileRes.do?declId=8555&amp;key=Publication-Content</li>\n<li>&quot;Improved Pareto/NBD Model and Its Applications in Customer Segmentation based on PErsonal Information Combination&quot; http://www.sersc.org/journals/IJDTA/vol6_no5/16.pdf</li>\n</ul>\n<p>In above NBD is negative binomial distribution which was used in early phases of computer modelling . See wiki page&nbsp;http://en.wikipedia.org/wiki/Negative_binomial_distribution<br>Remark that NBD modelling approach has flaws but it is better than nothing :)</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "116232",
      "postDate": "04/22/2016 15:54:48",
      "content": "<p>mark</p>",
      "rawMarkdown": "mark",
      "votes": null
    },
    {
      "id": "149245",
      "postDate": "12/08/2016 23:42:10",
      "content": "<p>How do i find accuracy of my model as I do not have test model results ???</p>",
      "rawMarkdown": "How do i find accuracy of my model as I do not have test model results ???",
      "votes": null
    },
    {
      "id": "152544",
      "postDate": "12/27/2016 03:06:18",
      "content": "<p>mark</p>",
      "rawMarkdown": "mark",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 116232,
      "author_name": "leyue260351",
      "author_url": "",
      "post_date": "04/22/2016 15:54:48",
      "content": "<p>mark</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 149245,
      "author_name": "shivayogi",
      "author_url": "",
      "post_date": "12/08/2016 23:42:10",
      "content": "<p>How do i find accuracy of my model as I do not have test model results ???</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 152544,
      "author_name": "sevenguin",
      "author_url": "",
      "post_date": "12/27/2016 03:06:18",
      "content": "<p>mark</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "41840": "",
    "116232": "mark",
    "149245": "How do i find accuracy of my model as I do not have test model results ???",
    "152544": "mark"
  },
  "source": "meta"
}