I noticed there are a lot of little chunks of data in the NombreProducto field. For example:
\n\nDalmata 1p 55g Tira MLA 523\nBollos BK 5in 20p 1730g TIR SL 135
\n\nDoes anyone know what all these chunks of letters and numbers mean?
\n\nObviously the first word is a description of the food. The in, p, g numbers are what I'm a bit uncertain of. Then what is the next 1 or two groupings of letters, some sort of special code? It looks like the last 3 numbers are just the product ID.
\n\nApparently the last group of letters is the brand (i.e. MLA).
", "messages": [ { "id": "130265", "postDate": "08/05/2016 05:38:53", "content": "I noticed there are a lot of little chunks of data in the NombreProducto field. For example:
\n\nDalmata 1p 55g Tira MLA 523\nBollos BK 5in 20p 1730g TIR SL 135
\n\nDoes anyone know what all these chunks of letters and numbers mean?
\n\nObviously the first word is a description of the food. The in, p, g numbers are what I'm a bit uncertain of. Then what is the next 1 or two groupings of letters, some sort of special code? It looks like the last 3 numbers are just the product ID.
\n\nApparently the last group of letters is the brand (i.e. MLA).
", "rawMarkdown": "I noticed there are a lot of little chunks of data in the NombreProducto field. For example:\r\n\r\nDalmata 1p 55g Tira MLA 523\r\nBollos BK 5in 20p 1730g TIR SL 135\r\n\r\nDoes anyone know what all these chunks of letters and numbers mean?\r\n\r\nObviously the first word is a description of the food. The in, p, g numbers are what I'm a bit uncertain of. Then what is the next 1 or two groupings of letters, some sort of special code? It looks like the last 3 numbers are just the product ID.\r\n\r\nApparently the [last group of letters is the brand][1] (i.e. MLA).\r\n\r\n\r\n [1]: https://www.kaggle.com/c/grupo-bimbo-inventory-demand/forums/t/22403/producto-tabla-csv/128464", "votes": null }, { "id": "130397", "postDate": "08/06/2016 16:02:08", "content": "Hi Nathan, you're right there is a lot of useful info in the NombreProducto field. \nFor example p or pq, means the pack size. Product ID: 30393 - "Powerade Naranja 12p 600ml CC 30393" is a box of 12 bottles of powerade of 600ml each. "CC" is the brand "coca cola"\nDepending on the brand you will find different unit of measurements ml, grams (g), kilograms (kg)\nAs you said the last part of the string is the product id same as "Product_ID"
\n\nGood luck\nReynaldo
", "rawMarkdown": "Hi Nathan, you're right there is a lot of useful info in the NombreProducto field. \r\nFor example p or pq, means the pack size. Product ID: 30393 - \"Powerade Naranja 12p 600ml CC 30393\" is a box of 12 bottles of powerade of 600ml each. \"CC\" is the brand \"coca cola\"\r\nDepending on the brand you will find different unit of measurements ml, grams (g), kilograms (kg)\r\nAs you said the last part of the string is the product id same as \"Product_ID\"\r\n\r\nGood luck\r\nReynaldo", "votes": null }, { "id": "130414", "postDate": "08/06/2016 19:35:21", "content": "I see, thanks. It would be nice to have a comprehensive list to know what exactly to look for, but this is helpful.
", "rawMarkdown": "I see, thanks. It would be nice to have a comprehensive list to know what exactly to look for, but this is helpful.", "votes": null }, { "id": "130468", "postDate": "08/07/2016 11:36:07", "content": "@NathanGeorges, you should read older posts in the forum. There is at least a couple of great notebooks exploiting product names to extract useful information.
", "rawMarkdown": "NathanGeorges, you should read older posts in the forum. There is at least a couple of great notebooks exploiting product names to extract useful information.", "votes": null }, { "id": "130493", "postDate": "08/07/2016 18:29:18", "content": "@CPMP Yes, I found a few:
\n\nThis one gets brand, weight, and pieces (Python).
\n\nThis one also gets brand, weight and pieces (R).
\n\nHere's another Python kernel that gets weight, volume, pieces, product, and brand.
\n\nAnother R kernel, for brand pieces and weight.
\n\nThis R kernel takes a bag-of-words approach.
\n\n\n\nI haven't made it through all the kernels because there are about 500.
\n\nI guess the definition of the product features doesn't really matter for machine learning, but it seems like it would be nice to know, in case we can generate more features from these product features (i.e. ratios).
\n\nSo far, I've come up with this as a set of features in each product name:
\n\ndesc, pct, other, inlabel, p, pq, g, ml, code1, code2, brand, p_id
\n\nI don't really know what the '150in' or '100pq' means.
\n\nI've made a kernel that extracts these features, although it still needs some improvement.
", "rawMarkdown": "CPMP Yes, I found a few:\r\n\r\n[This one gets brand, weight, and pieces (Python).][1]\r\n\r\n[This one also gets brand, weight and pieces (R).][2]\r\n\r\n[Here's another Python kernel that gets weight, volume, pieces, product, and brand.][3]\r\n\r\n[Another R kernel, for brand pieces and weight.][4]\r\n\r\n[This R kernel takes a bag-of-words approach.][5]\r\n\r\n[One more R kernel.][6]\r\n\r\nI haven't made it through all the kernels because there are about 500.\r\n\r\nI guess the definition of the product features doesn't really matter for machine learning, but it seems like it would be nice to know, in case we can generate more features from these product features (i.e. ratios).\r\n\r\nSo far, I've come up with this as a set of features in each product name:\r\n\r\ndesc, pct, other, inlabel, p, pq, g, ml, code1, code2, brand, p_id\r\n\r\nI don't really know what the '150in' or '100pq' means.\r\n\r\n[I've made a kernel that extracts these features, although it still needs some improvement.][7]\r\n\r\n\r\n [1]: https://www.kaggle.com/vykhand/grupo-bimbo-inventory-demand/exploring-products\r\n [2]: https://www.kaggle.com/lyytinen/grupo-bimbo-inventory-demand/basic-preprocessing-for-products\r\n [3]: https://www.kaggle.com/namra42/grupo-bimbo-inventory-demand/product-preprocessing\r\n [4]: https://www.kaggle.com/caiotaniguchi/grupo-bimbo-inventory-demand/basic-preprocessing-for-prods-modified\r\n [5]: https://www.kaggle.com/ybabakhin/grupo-bimbo-inventory-demand/products-clustering\r\n [6]: https://www.kaggle.com/evaxue/grupo-bimbo-inventory-demand/product-txt-clean-format/run/290653\r\n [7]: https://www.kaggle.com/wordsforthewise/grupo-bimbo-inventory-demand/extract-product-features", "votes": null } ], "comments": [ { "id": 130397, "author_name": "espanarey", "author_url": "", "post_date": "08/06/2016 16:02:08", "content": "Hi Nathan, you're right there is a lot of useful info in the NombreProducto field. \nFor example p or pq, means the pack size. Product ID: 30393 - "Powerade Naranja 12p 600ml CC 30393" is a box of 12 bottles of powerade of 600ml each. "CC" is the brand "coca cola"\nDepending on the brand you will find different unit of measurements ml, grams (g), kilograms (kg)\nAs you said the last part of the string is the product id same as "Product_ID"
\n\nGood luck\nReynaldo
", "votes": null, "replies": [] }, { "id": 130414, "author_name": "wordsforthewise", "author_url": "", "post_date": "08/06/2016 19:35:21", "content": "I see, thanks. It would be nice to have a comprehensive list to know what exactly to look for, but this is helpful.
", "votes": null, "replies": [] }, { "id": 130468, "author_name": "cpmpml", "author_url": "", "post_date": "08/07/2016 11:36:07", "content": "@NathanGeorges, you should read older posts in the forum. There is at least a couple of great notebooks exploiting product names to extract useful information.
", "votes": null, "replies": [] }, { "id": 130493, "author_name": "wordsforthewise", "author_url": "", "post_date": "08/07/2016 18:29:18", "content": "@CPMP Yes, I found a few:
\n\nThis one gets brand, weight, and pieces (Python).
\n\nThis one also gets brand, weight and pieces (R).
\n\nHere's another Python kernel that gets weight, volume, pieces, product, and brand.
\n\nAnother R kernel, for brand pieces and weight.
\n\nThis R kernel takes a bag-of-words approach.
\n\n\n\nI haven't made it through all the kernels because there are about 500.
\n\nI guess the definition of the product features doesn't really matter for machine learning, but it seems like it would be nice to know, in case we can generate more features from these product features (i.e. ratios).
\n\nSo far, I've come up with this as a set of features in each product name:
\n\ndesc, pct, other, inlabel, p, pq, g, ml, code1, code2, brand, p_id
\n\nI don't really know what the '150in' or '100pq' means.
\n\nI've made a kernel that extracts these features, although it still needs some improvement.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "130265": "I noticed there are a lot of little chunks of data in the NombreProducto field. For example:\r\n\r\nDalmata 1p 55g Tira MLA 523\r\nBollos BK 5in 20p 1730g TIR SL 135\r\n\r\nDoes anyone know what all these chunks of letters and numbers mean?\r\n\r\nObviously the first word is a description of the food. The in, p, g numbers are what I'm a bit uncertain of. Then what is the next 1 or two groupings of letters, some sort of special code? It looks like the last 3 numbers are just the product ID.\r\n\r\nApparently the [last group of letters is the brand][1] (i.e. MLA).\r\n\r\n\r\n [1]: https://www.kaggle.com/c/grupo-bimbo-inventory-demand/forums/t/22403/producto-tabla-csv/128464", "130397": "Hi Nathan, you're right there is a lot of useful info in the NombreProducto field. \r\nFor example p or pq, means the pack size. Product ID: 30393 - \"Powerade Naranja 12p 600ml CC 30393\" is a box of 12 bottles of powerade of 600ml each. \"CC\" is the brand \"coca cola\"\r\nDepending on the brand you will find different unit of measurements ml, grams (g), kilograms (kg)\r\nAs you said the last part of the string is the product id same as \"Product_ID\"\r\n\r\nGood luck\r\nReynaldo", "130414": "I see, thanks. It would be nice to have a comprehensive list to know what exactly to look for, but this is helpful.", "130468": "NathanGeorges, you should read older posts in the forum. There is at least a couple of great notebooks exploiting product names to extract useful information.", "130493": "CPMP Yes, I found a few:\r\n\r\n[This one gets brand, weight, and pieces (Python).][1]\r\n\r\n[This one also gets brand, weight and pieces (R).][2]\r\n\r\n[Here's another Python kernel that gets weight, volume, pieces, product, and brand.][3]\r\n\r\n[Another R kernel, for brand pieces and weight.][4]\r\n\r\n[This R kernel takes a bag-of-words approach.][5]\r\n\r\n[One more R kernel.][6]\r\n\r\nI haven't made it through all the kernels because there are about 500.\r\n\r\nI guess the definition of the product features doesn't really matter for machine learning, but it seems like it would be nice to know, in case we can generate more features from these product features (i.e. ratios).\r\n\r\nSo far, I've come up with this as a set of features in each product name:\r\n\r\ndesc, pct, other, inlabel, p, pq, g, ml, code1, code2, brand, p_id\r\n\r\nI don't really know what the '150in' or '100pq' means.\r\n\r\n[I've made a kernel that extracts these features, although it still needs some improvement.][7]\r\n\r\n\r\n [1]: https://www.kaggle.com/vykhand/grupo-bimbo-inventory-demand/exploring-products\r\n [2]: https://www.kaggle.com/lyytinen/grupo-bimbo-inventory-demand/basic-preprocessing-for-products\r\n [3]: https://www.kaggle.com/namra42/grupo-bimbo-inventory-demand/product-preprocessing\r\n [4]: https://www.kaggle.com/caiotaniguchi/grupo-bimbo-inventory-demand/basic-preprocessing-for-prods-modified\r\n [5]: https://www.kaggle.com/ybabakhin/grupo-bimbo-inventory-demand/products-clustering\r\n [6]: https://www.kaggle.com/evaxue/grupo-bimbo-inventory-demand/product-txt-clean-format/run/290653\r\n [7]: https://www.kaggle.com/wordsforthewise/grupo-bimbo-inventory-demand/extract-product-features" }, "source": "meta" }