{ "id": 306858, "title": "Know Your Customer Before Recommending 👀✔️❌", "url": "/competitions/h-and-m-personalized-fashion-recommendations/discussion/306858", "author_name": "", "post_date": "2022-02-11T08:26:33.426361200Z", "votes": 5, "comment_count": 5, "views": 0, "content": "

It is important to know the customer behavior before recommending. My thoughts are to find a cluster based on customer Recency, Frequency and Monetary value aka RFM and know

\n", "messages": [ { "id": "1685393", "postDate": "02/11/2022 08:26:33", "content": "

It is important to know the customer behavior before recommending. My thoughts are to find a cluster based on customer Recency, Frequency and Monetary value aka RFM and know

\n", "rawMarkdown": "It is important to know the customer behavior before recommending. My thoughts are to find a cluster based on customer Recency, Frequency and Monetary value aka RFM and know \n- 1) Most Valued Customer \n- 2) Lazy Customer \n- 3) Fashion Freak \n- 4) Frequent Customer etc.\nIs the customer purchase behavior changed over the year? Due to current situation of Pandemic does customer moved from offline to online purchase. The customer understanding will improve the final recommendation. I will share the notebook and the output soon. Please let me know your comments.", "votes": null }, { "id": "1685744", "postDate": "02/11/2022 14:12:25", "content": "

An alternative is just to do feature engineering and add features that capture this information.

\n

For example, for #4, you could create a simple aggregation feature that has how many times a customer purchased, how many times in the last year etc.

", "rawMarkdown": "An alternative is just to do feature engineering and add features that capture this information.\n\nFor example, for #4, you could create a simple aggregation feature that has how many times a customer purchased, how many times in the last year etc.", "votes": null }, { "id": "1686020", "postDate": "02/11/2022 17:39:46", "content": "

True. Feature engineering will give you some of the information but RFM is for customer groups. Let's see how the pareto chart of customer purchase pattern looks like.

", "rawMarkdown": "True. Feature engineering will give you some of the information but RFM is for customer groups. Let's see how the pareto chart of customer purchase pattern looks like.", "votes": null }, { "id": "1689331", "postDate": "02/14/2022 06:40:06", "content": "

I did some digging into how often customers purchase within the two-year period. I found that the number of items purchased ranging from 1 to 1895, with median equal to 9 and a mean of 23.3.

", "rawMarkdown": "I did some digging into how often customers purchase within the two-year period. I found that the number of items purchased ranging from 1 to 1895, with median equal to 9 and a mean of 23.3.", "votes": null }, { "id": "1696963", "postDate": "02/19/2022 09:06:20", "content": "

I think,considering customer age as one of the clustering factors can also provide valuable insights along with RFM.

", "rawMarkdown": "I think,considering customer age as one of the clustering factors can also provide valuable insights along with RFM.", "votes": null }, { "id": "1696964", "postDate": "02/19/2022 09:09:13", "content": "

good work. will look into it more deeply

", "rawMarkdown": "good work. will look into it more deeply", "votes": null } ], "comments": [ { "id": 1685744, "author_name": "jacob34", "author_url": "", "post_date": "02/11/2022 14:12:25", "content": "

An alternative is just to do feature engineering and add features that capture this information.

\n

For example, for #4, you could create a simple aggregation feature that has how many times a customer purchased, how many times in the last year etc.

", "votes": null, "replies": [] }, { "id": 1686020, "author_name": "pankajkumar", "author_url": "", "post_date": "02/11/2022 17:39:46", "content": "

True. Feature engineering will give you some of the information but RFM is for customer groups. Let's see how the pareto chart of customer purchase pattern looks like.

", "votes": null, "replies": [] }, { "id": 1689331, "author_name": "joechan", "author_url": "", "post_date": "02/14/2022 06:40:06", "content": "

I did some digging into how often customers purchase within the two-year period. I found that the number of items purchased ranging from 1 to 1895, with median equal to 9 and a mean of 23.3.

", "votes": null, "replies": [] }, { "id": 1696963, "author_name": "sameersanjaybhosale", "author_url": "", "post_date": "02/19/2022 09:06:20", "content": "

I think,considering customer age as one of the clustering factors can also provide valuable insights along with RFM.

", "votes": null, "replies": [] }, { "id": 1696964, "author_name": "beraamit", "author_url": "", "post_date": "02/19/2022 09:09:13", "content": "

good work. will look into it more deeply

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1685393": "It is important to know the customer behavior before recommending. My thoughts are to find a cluster based on customer Recency, Frequency and Monetary value aka RFM and know \n- 1) Most Valued Customer \n- 2) Lazy Customer \n- 3) Fashion Freak \n- 4) Frequent Customer etc.\nIs the customer purchase behavior changed over the year? Due to current situation of Pandemic does customer moved from offline to online purchase. The customer understanding will improve the final recommendation. I will share the notebook and the output soon. Please let me know your comments.", "1685744": "An alternative is just to do feature engineering and add features that capture this information.\n\nFor example, for #4, you could create a simple aggregation feature that has how many times a customer purchased, how many times in the last year etc.", "1686020": "True. Feature engineering will give you some of the information but RFM is for customer groups. Let's see how the pareto chart of customer purchase pattern looks like.", "1689331": "I did some digging into how often customers purchase within the two-year period. I found that the number of items purchased ranging from 1 to 1895, with median equal to 9 and a mean of 23.3.", "1696963": "I think,considering customer age as one of the clustering factors can also provide valuable insights along with RFM.", "1696964": "good work. will look into it more deeply" }, "source": "meta" }