{
  "id": 78574,
  "title": "What does loyalty score has to do with Elo's objective?",
  "url": "/competitions/elo-merchant-category-recommendation/discussion/78574",
  "author_name": "",
  "post_date": "2019-01-25T13:27:51.598912700Z",
  "votes": 4,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>I know there has been lots of discussions about the loyalty score and how it's related to the give datasets. But I don't understand how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?</p>\n\n<p>Regardless of how the loyalty score is calculated and how it's related to the given data points, a loyalty score for every card_id with respect to every merchant_id is what would make sense to me! In other words, I might be a loyal customer for Uber, but no so loyal to McDonalds! How would I be evaluated in that case? Are they associating the highest loyalty score to every card_id? or the average? or I am missing something here!?</p>\n\n<p>In any case, how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!? </p>",
  "messages": [
    {
      "id": "461182",
      "postDate": "01/25/2019 13:27:51",
      "content": "<p>Hi,</p>\n\n<p>I know there has been lots of discussions about the loyalty score and how it's related to the give datasets. But I don't understand how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?</p>\n\n<p>Regardless of how the loyalty score is calculated and how it's related to the given data points, a loyalty score for every card_id with respect to every merchant_id is what would make sense to me! In other words, I might be a loyal customer for Uber, but no so loyal to McDonalds! How would I be evaluated in that case? Are they associating the highest loyalty score to every card_id? or the average? or I am missing something here!?</p>\n\n<p>In any case, how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!? </p>",
      "rawMarkdown": "Hi,\n\nI know there has been lots of discussions about the loyalty score and how it's related to the give datasets. But I don't understand how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?\n\nRegardless of how the loyalty score is calculated and how it's related to the given data points, a loyalty score for every card_id with respect to every merchant_id is what would make sense to me! In other words, I might be a loyal customer for Uber, but no so loyal to McDonalds! How would I be evaluated in that case? Are they associating the highest loyalty score to every card_id? or the average? or I am missing something here!?\n\nIn any case, how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?",
      "votes": null
    },
    {
      "id": "461288",
      "postDate": "01/25/2019 17:50:08",
      "content": "<p>I agree with you completely that a loyalty score for each card with respect to each merchant would be a much more interesting and useful problem. That being said I think they're trying to determine if a customer is <em>overall</em> loyal, relative to other customers or in absolute terms. If there's a customer they know to be loyal, they'd avoid sending them coupons to businesses that are competitors of businesses that they already frequent, and they'd send them discounts for businesses in categories they don't already transact in. This way they could prevent wasting coupons on someone who won't leave their current brand, and would ensure that high value opportunities (a high-loyalty customer who doesn't have, say... a preferred supermarket brand) could be targeted for discounts and advertisements. </p>",
      "rawMarkdown": "I agree with you completely that a loyalty score for each card with respect to each merchant would be a much more interesting and useful problem. That being said I think they're trying to determine if a customer is *overall* loyal, relative to other customers or in absolute terms. If there's a customer they know to be loyal, they'd avoid sending them coupons to businesses that are competitors of businesses that they already frequent, and they'd send them discounts for businesses in categories they don't already transact in. This way they could prevent wasting coupons on someone who won't leave their current brand, and would ensure that high value opportunities (a high-loyalty customer who doesn't have, say... a preferred supermarket brand) could be targeted for discounts and advertisements.",
      "votes": null
    },
    {
      "id": "461647",
      "postDate": "01/26/2019 16:28:03",
      "content": "<p>I've said this a couple times in the forums, and I'm not sure I'm correct, but it seems to me that Elo knows the loyalty score for each card (whatever it is based on), and they have transaction histories for each card (which the loyalty score is <em>not</em> based on).  Now they want to know if the loyalty score can be predicted by examining the transaction history.</p>",
      "rawMarkdown": "I've said this a couple times in the forums, and I'm not sure I'm correct, but it seems to me that Elo knows the loyalty score for each card (whatever it is based on), and they have transaction histories for each card (which the loyalty score is *not* based on).  Now they want to know if the loyalty score can be predicted by examining the transaction history.",
      "votes": null
    },
    {
      "id": "461658",
      "postDate": "01/26/2019 16:42:36",
      "content": "<p>Thanks Michael, that seems to be the popular theory here. But I must say, that this does not add up with what they say in the Overview:</p>\n\n<blockquote>\n  <p>In this competition, Kagglers will develop algorithms to identify and serve the most relevant opportunities to individuals, by uncovering signal in customer loyalty. Your input will improve customers’ lives and help Elo reduce unwanted campaigns, to create the right experience for customers.</p>\n</blockquote>",
      "rawMarkdown": "Thanks Michael, that seems to be the popular theory here. But I must say, that this does not add up with what they say in the Overview:\n&gt;In this competition, Kagglers will develop algorithms to identify and serve the most relevant opportunities to individuals, by uncovering signal in customer loyalty. Your input will improve customers’ lives and help Elo reduce unwanted campaigns, to create the right experience for customers.",
      "votes": null
    },
    {
      "id": "462260",
      "postDate": "01/28/2019 01:32:05",
      "content": "<p>as Michael Maguire mentioned, if Elo is trying to predict loyalty score which is not based on transaction data, It makes sense. Assuming hidden consumer behavior patterns in transaction data does leads you to predict any kinds of loyalty, it means that pattern that is very unique to individual is what to be handled in the promotions or recommendation in the future of Elo marketing. So the point is that Elo just want to find out hidden behavior pattern that affects their loyalty score so they can handle promotions in individual levels. of course this is just my opinion</p>",
      "rawMarkdown": "as Michael Maguire mentioned, if Elo is trying to predict loyalty score which is not based on transaction data, It makes sense. Assuming hidden consumer behavior patterns in transaction data does leads you to predict any kinds of loyalty, it means that pattern that is very unique to individual is what to be handled in the promotions or recommendation in the future of Elo marketing. So the point is that Elo just want to find out hidden behavior pattern that affects their loyalty score so they can handle promotions in individual levels. of course this is just my opinion",
      "votes": null
    },
    {
      "id": "462347",
      "postDate": "01/28/2019 06:43:33",
      "content": "<p>Actually for Predicting loyalty we need Coustmer transactions behaviour So we need transactional Data of that Customer.\nLoyalty and Discount is based on \nHow much regular Customer is he/she so we need Time data. </p>",
      "rawMarkdown": "Actually for Predicting loyalty we need Coustmer transactions behaviour So we need transactional Data of that Customer.\nLoyalty and Discount is based on \nHow much regular Customer is he/she so we need Time data.",
      "votes": null
    },
    {
      "id": "462546",
      "postDate": "01/28/2019 13:34:40",
      "content": "<p>Good Thnx</p>",
      "rawMarkdown": "Good Thnx",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 461288,
      "author_name": "alecthekulak",
      "author_url": "",
      "post_date": "01/25/2019 17:50:08",
      "content": "<p>I agree with you completely that a loyalty score for each card with respect to each merchant would be a much more interesting and useful problem. That being said I think they're trying to determine if a customer is <em>overall</em> loyal, relative to other customers or in absolute terms. If there's a customer they know to be loyal, they'd avoid sending them coupons to businesses that are competitors of businesses that they already frequent, and they'd send them discounts for businesses in categories they don't already transact in. This way they could prevent wasting coupons on someone who won't leave their current brand, and would ensure that high value opportunities (a high-loyalty customer who doesn't have, say... a preferred supermarket brand) could be targeted for discounts and advertisements. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 461647,
      "author_name": "thawatt",
      "author_url": "",
      "post_date": "01/26/2019 16:28:03",
      "content": "<p>I've said this a couple times in the forums, and I'm not sure I'm correct, but it seems to me that Elo knows the loyalty score for each card (whatever it is based on), and they have transaction histories for each card (which the loyalty score is <em>not</em> based on).  Now they want to know if the loyalty score can be predicted by examining the transaction history.</p>",
      "votes": null,
      "replies": [
        {
          "id": 461658,
          "author_name": "hoss3in",
          "author_url": "",
          "post_date": "01/26/2019 16:42:36",
          "content": "<p>Thanks Michael, that seems to be the popular theory here. But I must say, that this does not add up with what they say in the Overview:</p>\n\n<blockquote>\n  <p>In this competition, Kagglers will develop algorithms to identify and serve the most relevant opportunities to individuals, by uncovering signal in customer loyalty. Your input will improve customers’ lives and help Elo reduce unwanted campaigns, to create the right experience for customers.</p>\n</blockquote>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 462260,
      "author_name": "khajun",
      "author_url": "",
      "post_date": "01/28/2019 01:32:05",
      "content": "<p>as Michael Maguire mentioned, if Elo is trying to predict loyalty score which is not based on transaction data, It makes sense. Assuming hidden consumer behavior patterns in transaction data does leads you to predict any kinds of loyalty, it means that pattern that is very unique to individual is what to be handled in the promotions or recommendation in the future of Elo marketing. So the point is that Elo just want to find out hidden behavior pattern that affects their loyalty score so they can handle promotions in individual levels. of course this is just my opinion</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 462347,
      "author_name": "tahirmuslimscientist",
      "author_url": "",
      "post_date": "01/28/2019 06:43:33",
      "content": "<p>Actually for Predicting loyalty we need Coustmer transactions behaviour So we need transactional Data of that Customer.\nLoyalty and Discount is based on \nHow much regular Customer is he/she so we need Time data. </p>",
      "votes": null,
      "replies": [
        {
          "id": 462546,
          "author_name": "tahirsoft",
          "author_url": "",
          "post_date": "01/28/2019 13:34:40",
          "content": "<p>Good Thnx</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "461182": "Hi,\n\nI know there has been lots of discussions about the loyalty score and how it's related to the give datasets. But I don't understand how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?\n\nRegardless of how the loyalty score is calculated and how it's related to the given data points, a loyalty score for every card_id with respect to every merchant_id is what would make sense to me! In other words, I might be a loyal customer for Uber, but no so loyal to McDonalds! How would I be evaluated in that case? Are they associating the highest loyalty score to every card_id? or the average? or I am missing something here!?\n\nIn any case, how a single loyalty score for every card_id would serve Elo to tailor discount offers for every card!?",
    "461288": "I agree with you completely that a loyalty score for each card with respect to each merchant would be a much more interesting and useful problem. That being said I think they're trying to determine if a customer is *overall* loyal, relative to other customers or in absolute terms. If there's a customer they know to be loyal, they'd avoid sending them coupons to businesses that are competitors of businesses that they already frequent, and they'd send them discounts for businesses in categories they don't already transact in. This way they could prevent wasting coupons on someone who won't leave their current brand, and would ensure that high value opportunities (a high-loyalty customer who doesn't have, say... a preferred supermarket brand) could be targeted for discounts and advertisements.",
    "461647": "I've said this a couple times in the forums, and I'm not sure I'm correct, but it seems to me that Elo knows the loyalty score for each card (whatever it is based on), and they have transaction histories for each card (which the loyalty score is *not* based on).  Now they want to know if the loyalty score can be predicted by examining the transaction history.",
    "461658": "Thanks Michael, that seems to be the popular theory here. But I must say, that this does not add up with what they say in the Overview:\n&gt;In this competition, Kagglers will develop algorithms to identify and serve the most relevant opportunities to individuals, by uncovering signal in customer loyalty. Your input will improve customers’ lives and help Elo reduce unwanted campaigns, to create the right experience for customers.",
    "462260": "as Michael Maguire mentioned, if Elo is trying to predict loyalty score which is not based on transaction data, It makes sense. Assuming hidden consumer behavior patterns in transaction data does leads you to predict any kinds of loyalty, it means that pattern that is very unique to individual is what to be handled in the promotions or recommendation in the future of Elo marketing. So the point is that Elo just want to find out hidden behavior pattern that affects their loyalty score so they can handle promotions in individual levels. of course this is just my opinion",
    "462347": "Actually for Predicting loyalty we need Coustmer transactions behaviour So we need transactional Data of that Customer.\nLoyalty and Discount is based on \nHow much regular Customer is he/she so we need Time data.",
    "462546": "Good Thnx"
  },
  "source": "meta"
}