This Google study presents initial results on predicting real-life user experience (UX) on laptops from their hardware specifications. The research focuses on web applications that run on Chromebooks. By tracking 9 UX metrics on common end-user workloads such as web browsing, video playback, and audio/video calls, the study uses a dataset of 100K UX data points to train gradient boosted regression trees that predict metric values from device specifications. The findings show a high accuracy in the predictions, with a mean R2 score of 97.8% and a mean MAAPE of 10.1%.
Publication date: 14 Feb 2024
Project Page: https://arxiv.org/abs/2402.08964v1
Paper: https://arxiv.org/pdf/2402.08964
