Uncovering and Understanding Smart TV Users’ Picture Quality Preferences via Big Data Analytics
Hosub Lee · 2023
We analyzed one year’s device usage log data collected from Smart TVs launched in the US to uncover and understand user-preferred picture quality previously unknown. Through K-means cluster analysis, we first clustered similar users into groups based on their historical picture setting activities on TVs. We then performed K-prototypes clustering upon data instances from a specific group of users, who were reasonably active in customizing picture quality, to discover new picture setting configurations distinguished from the factory default. Lastly, we conducted mixed effects modeling on the collected datasets to determine contextual factors impacting the user preferences towards the discovered picture settings. The analysis yielded several implications for the consumer electronics industry in providing users with the picture quality best suited for their current situation and personal characteristics.