Hi, concerning this statement : \"You will be making purchase predictions for all customer_id values provided, regardless of whether these customers made purchases in the training data\".
\nI am quite confused here because as far as I could explore the only information we have about the customer is his purchasing history. Without this info i dont see how can we accurately recommend articles.
\nAny advice here ?.
\nBest;
Hi, concerning this statement : \"You will be making purchase predictions for all customer_id values provided, regardless of whether these customers made purchases in the training data\".
\nI am quite confused here because as far as I could explore the only information we have about the customer is his purchasing history. Without this info i dont see how can we accurately recommend articles.
\nAny advice here ?.
\nBest;
Hi Jose,
\nit is not an advice, but we can use the popular articles.
Other way is to get the pattern of new user of each week on what did they buy.
", "rawMarkdown": "Hi Jose, \nit is not an advice, but we can use the popular articles.\n\nOther way is to get the pattern of new user of each week on what did they buy.", "votes": null }, { "id": "1722583", "postDate": "03/14/2022 17:19:23", "content": "There are also a few other features in the customers csv file, which we can use to group a customer without purchasing history with other customers who have purchasing history.
", "rawMarkdown": "There are also a few other features in the customers csv file, which we can use to group a customer without purchasing history with other customers who have purchasing history.", "votes": null }, { "id": "1722695", "postDate": "03/14/2022 19:33:16", "content": "Ok thanks for these tips, first time in such a competition.
", "rawMarkdown": "Ok thanks for these tips, first time in such a competition.", "votes": null }, { "id": "1722697", "postDate": "03/14/2022 19:34:00", "content": "thanks for the tip.
", "rawMarkdown": "thanks for the tip.", "votes": null }, { "id": "1722782", "postDate": "03/14/2022 21:43:03", "content": "You have info about customers like age, postal code, and other features, no matter if it is the transactions history or not. So you can use this data to match with similar customers that have transactions and get some predictions.
", "rawMarkdown": "You have info about customers like age, postal code, and other features, no matter if it is the transactions history or not. So you can use this data to match with similar customers that have transactions and get some predictions.", "votes": null }, { "id": "1722801", "postDate": "03/14/2022 21:53:49", "content": "Probably wouldn't matter either way. For somebody who hasn't bought anything in 2 years, you might as well as put dummy predictions and not have your score affected.
", "rawMarkdown": "Probably wouldn't matter either way. For somebody who hasn't bought anything in 2 years, you might as well as put dummy predictions and not have your score affected.", "votes": null }, { "id": "1749357", "postDate": "04/08/2022 13:56:42", "content": "Question is - if a customer_id is not present in the training set (e.g. 7034c909b5f39e6b184fe36b9edda3bf62b26509222ad6eb2744b3e1f8c02c2e), then do we know for certain that they will make a purchase in the test set period?
Because if not, then it seems odd that they'd give us customer ids for customer who don't make a prediction neither during training nor during test. But if we do know for certain that they will make a purchase, then that seems like quite a lot of information to give contestants, and not what one would typically have available to them in a real-life data science task
", "rawMarkdown": "Question is - if a `customer_id` is not present in the training set (e.g. `7034c909b5f39e6b184fe36b9edda3bf62b26509222ad6eb2744b3e1f8c02c2e`), then do we know for certain that they will make a purchase in the test set period?\n\nBecause if not, then it seems odd that they'd give us customer ids for customer who don't make a prediction neither during training nor during test. But if we do know for certain that they will make a purchase, then that seems like quite a lot of information to give contestants, and not what one would typically have available to them in a real-life data science task", "votes": null } ], "comments": [ { "id": 1722561, "author_name": "hervind", "author_url": "", "post_date": "03/14/2022 16:49:06", "content": "Hi Jose,
\nit is not an advice, but we can use the popular articles.
Other way is to get the pattern of new user of each week on what did they buy.
", "votes": null, "replies": [ { "id": 1722695, "author_name": "jsaray", "author_url": "", "post_date": "03/14/2022 19:33:16", "content": "Ok thanks for these tips, first time in such a competition.
", "votes": null, "replies": [] } ] }, { "id": 1722583, "author_name": "jacob34", "author_url": "", "post_date": "03/14/2022 17:19:23", "content": "There are also a few other features in the customers csv file, which we can use to group a customer without purchasing history with other customers who have purchasing history.
", "votes": null, "replies": [ { "id": 1722697, "author_name": "jsaray", "author_url": "", "post_date": "03/14/2022 19:34:00", "content": "thanks for the tip.
", "votes": null, "replies": [] } ] }, { "id": 1722782, "author_name": "ginkobalboa", "author_url": "", "post_date": "03/14/2022 21:43:03", "content": "You have info about customers like age, postal code, and other features, no matter if it is the transactions history or not. So you can use this data to match with similar customers that have transactions and get some predictions.
", "votes": null, "replies": [] }, { "id": 1722801, "author_name": "atulverma", "author_url": "", "post_date": "03/14/2022 21:53:49", "content": "Probably wouldn't matter either way. For somebody who hasn't bought anything in 2 years, you might as well as put dummy predictions and not have your score affected.
", "votes": null, "replies": [] }, { "id": 1749357, "author_name": "marcogorelli", "author_url": "", "post_date": "04/08/2022 13:56:42", "content": "Question is - if a customer_id is not present in the training set (e.g. 7034c909b5f39e6b184fe36b9edda3bf62b26509222ad6eb2744b3e1f8c02c2e), then do we know for certain that they will make a purchase in the test set period?
Because if not, then it seems odd that they'd give us customer ids for customer who don't make a prediction neither during training nor during test. But if we do know for certain that they will make a purchase, then that seems like quite a lot of information to give contestants, and not what one would typically have available to them in a real-life data science task
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1722535": "Hi, concerning this statement : \"You will be making purchase predictions for all customer_id values provided, regardless of whether these customers made purchases in the training data\".\n\nI am quite confused here because as far as I could explore the only information we have about the customer is his purchasing history. Without this info i dont see how can we accurately recommend articles.\n\nAny advice here ?.\nBest;", "1722561": "Hi Jose, \nit is not an advice, but we can use the popular articles.\n\nOther way is to get the pattern of new user of each week on what did they buy.", "1722583": "There are also a few other features in the customers csv file, which we can use to group a customer without purchasing history with other customers who have purchasing history.", "1722695": "Ok thanks for these tips, first time in such a competition.", "1722697": "thanks for the tip.", "1722782": "You have info about customers like age, postal code, and other features, no matter if it is the transactions history or not. So you can use this data to match with similar customers that have transactions and get some predictions.", "1722801": "Probably wouldn't matter either way. For somebody who hasn't bought anything in 2 years, you might as well as put dummy predictions and not have your score affected.", "1749357": "Question is - if a `customer_id` is not present in the training set (e.g. `7034c909b5f39e6b184fe36b9edda3bf62b26509222ad6eb2744b3e1f8c02c2e`), then do we know for certain that they will make a purchase in the test set period?\n\nBecause if not, then it seems odd that they'd give us customer ids for customer who don't make a prediction neither during training nor during test. But if we do know for certain that they will make a purchase, then that seems like quite a lot of information to give contestants, and not what one would typically have available to them in a real-life data science task" }, "source": "meta" }