Shopping list generation with machine learning
Abstract: When households do their grocery shopping some sort of shopping list is used to make the shopping easier. The lists contain the groceries that are intended for purchase. These lists can be boring to make and is also not free from errors, so an automated way to generate these lists would be practical. This Master’s thesis aims to generate these shopping lists with data from past grocery receipts by predicting future receipts. We classified the groceries on the receipts into categories that are organized into two layers of categories, 209 subcategories and 17 main categories. These categories are modeled as time series with an indicator variable that models purchase/no purchase in the category. This indicator variable is estimated by using linear support vector machines in combination with an intensity expectation. The quantity of the groceries uses a Gaussian field as a model and is estimated with ordinary kriging. The data contains 15,969 groceries, from 1,230 receipts and 34 households. The quantity of a grocery on a receipt is measured by using the price paid for the item.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)