•Univariate regression was applied on the recently released ASHRAE database.•Individual, building, and geographic factors have been examined.•Geographic and HVAC mode are the most influential for neutral temperature.•Building type and local climate has the biggest impact on thermal sensitivity.•This work could help select useful features for personal comfort models. Different thermal demands and preferences between individuals lead to a low occupant satisfaction rate, despite the high energy consumption by HVAC system. This study aims to quantify the difference in thermal demands, and to compare the influential factors which might lead to those differences. With the recently released ASHRAE Database, we quantitatively answered the following two research questions: which factors would lead to marked individual difference, and what the magnitude of this difference is. Linear regression has been applied to describe the macro-trend of how people feel thermally under different temperatures. Three types of factors which might lead to different thermal demands have been studied and compared in this study, i.e. individual factors, building characteristics and geographical factors. It was found that the local climate has the most marked impact on the neutral temperature, with an effect size of 3.5 °C; followed by country, HVAC operation mode and body built, which lead to a difference of more than 1 °C. In terms of the thermal sensitivity, building type and local climate are the most influential factors. Subjects in residential buildings or coming from Dry climate zone could accept 2.5 °C wider temperature range than those in office, education buildings or from Continental climate zone. The findings of this research could help thermal comfort researchers and designers to identify influential factors that might lead to individual difference, and could shed light on the feature selection for the development of personal comfort models.