Educational project
Predicting the Fabric Consumption of Fashion Items
SUPSI Image Focus
We develop a machine learning model to predict the amount of fabric required by a new fashion item. In collaboration with a fashion company, we show that the adoption of the models can significantly reduce the production waste. The thesis won the Excellence award by Fondazione Nizzola.
When developing new fashion item, a correct prediction of the needed fabric allows to buy a correct stock of fabric, minimizing waste while allowing the production of the desired items. We collaborated with a fashion company, developing a machine learning model that predicts fabric consumption for a wide variety of items, considering both their technical characteristic and their design.
While baseline models are not precise enough to satisfy the error requirements of our partner, our customized Extreme Gradient Boosting model generates prediction which are accurate enough to be used to in planning and production.
While baseline models are not precise enough to satisfy the error requirements of our partner, our customized Extreme Gradient Boosting model generates prediction which are accurate enough to be used to in planning and production.