hi I went through the problem many times but still i cant understand the problem itself, please help me to understand ,from my understanding the elo wants to suggest its customer (card holders) near by shop (which may have discounts),so here given card holders data and merchants data ,we should be able predict whether the customer will buy from the certain merchant or not. but here we are trying to find the loyalty score of the card holder, which is from my understanding if elo suggest something whether the customer will do it or not? . I dont get the relationship between these things. I dont get why we even predicting loyalty score ?
", "messages": [ { "id": "460504", "postDate": "01/23/2019 21:20:14", "content": "hi I went through the problem many times but still i cant understand the problem itself, please help me to understand ,from my understanding the elo wants to suggest its customer (card holders) near by shop (which may have discounts),so here given card holders data and merchants data ,we should be able predict whether the customer will buy from the certain merchant or not. but here we are trying to find the loyalty score of the card holder, which is from my understanding if elo suggest something whether the customer will do it or not? . I dont get the relationship between these things. I dont get why we even predicting loyalty score ?
", "rawMarkdown": "hi I went through the problem many times but still i cant understand the problem itself, please help me to understand ,from my understanding the elo wants to suggest its customer (card holders) near by shop (which may have discounts),so here given card holders data and merchants data ,we should be able predict whether the customer will buy from the certain merchant or not. but here we are trying to find the loyalty score of the card holder, which is from my understanding if elo suggest something whether the customer will do it or not? . I dont get the relationship between these things. I dont get why we even predicting loyalty score ?", "votes": null }, { "id": "460558", "postDate": "01/24/2019 00:30:23", "content": "You are trying to predict the \"target\" column using the data in all the other columns (and datasets, since they were designed to be joined). As the description said, this is not based on any real data, the card IDs are just unique indicators for each row (which allows you to join the data from multiple tables). The goal is to take all of this data and predict the loyalty score that you (your model) believes is most likely to result. Some of the datasets include more rows than there are unique card IDs, and for those, feature engineering may be necessary to properly incorporate this data into your model.
\n\nSince all the data is simulated, there's not much underlying it to \"explain\" the problem in classical terms, however, the ultimate goal is likely to evaluate the effect of a number of independent variables on a consumer's loyalty to a business. This can be applied out-of-sample to determine the likelihood that a consumer will be loyal to a business they have never patronized before. This \"predicted loyalty\" could be used to help decide which businesses/restaurants to recommend to a consumer.
", "rawMarkdown": "You are trying to predict the \"target\" column using the data in all the other columns (and datasets, since they were designed to be joined). As the description said, this is not based on any real data, the card IDs are just unique indicators for each row (which allows you to join the data from multiple tables). The goal is to take all of this data and predict the loyalty score that you (your model) believes is most likely to result. Some of the datasets include more rows than there are unique card IDs, and for those, feature engineering may be necessary to properly incorporate this data into your model. \n\nSince all the data is simulated, there's not much underlying it to \"explain\" the problem in classical terms, however, the ultimate goal is likely to evaluate the effect of a number of independent variables on a consumer's loyalty to a business. This can be applied out-of-sample to determine the likelihood that a consumer will be loyal to a business they have never patronized before. This \"predicted loyalty\" could be used to help decide which businesses/restaurants to recommend to a consumer.", "votes": null } ], "comments": [ { "id": 460558, "author_name": "alecthekulak", "author_url": "", "post_date": "01/24/2019 00:30:23", "content": "You are trying to predict the \"target\" column using the data in all the other columns (and datasets, since they were designed to be joined). As the description said, this is not based on any real data, the card IDs are just unique indicators for each row (which allows you to join the data from multiple tables). The goal is to take all of this data and predict the loyalty score that you (your model) believes is most likely to result. Some of the datasets include more rows than there are unique card IDs, and for those, feature engineering may be necessary to properly incorporate this data into your model.
\n\nSince all the data is simulated, there's not much underlying it to \"explain\" the problem in classical terms, however, the ultimate goal is likely to evaluate the effect of a number of independent variables on a consumer's loyalty to a business. This can be applied out-of-sample to determine the likelihood that a consumer will be loyal to a business they have never patronized before. This \"predicted loyalty\" could be used to help decide which businesses/restaurants to recommend to a consumer.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "460504": "hi I went through the problem many times but still i cant understand the problem itself, please help me to understand ,from my understanding the elo wants to suggest its customer (card holders) near by shop (which may have discounts),so here given card holders data and merchants data ,we should be able predict whether the customer will buy from the certain merchant or not. but here we are trying to find the loyalty score of the card holder, which is from my understanding if elo suggest something whether the customer will do it or not? . I dont get the relationship between these things. I dont get why we even predicting loyalty score ?", "460558": "You are trying to predict the \"target\" column using the data in all the other columns (and datasets, since they were designed to be joined). As the description said, this is not based on any real data, the card IDs are just unique indicators for each row (which allows you to join the data from multiple tables). The goal is to take all of this data and predict the loyalty score that you (your model) believes is most likely to result. Some of the datasets include more rows than there are unique card IDs, and for those, feature engineering may be necessary to properly incorporate this data into your model. \n\nSince all the data is simulated, there's not much underlying it to \"explain\" the problem in classical terms, however, the ultimate goal is likely to evaluate the effect of a number of independent variables on a consumer's loyalty to a business. This can be applied out-of-sample to determine the likelihood that a consumer will be loyal to a business they have never patronized before. This \"predicted loyalty\" could be used to help decide which businesses/restaurants to recommend to a consumer." }, "source": "meta" }