Basic methods used for data analysis in adaptive thermal comfort studies

Abstract

Existing research literature on adaptive thermal comfort studies shows the methodology used and results obtained; however, the information for data analysis is reduced. Methods commonly used to interpret and understand the thermal comfort phenomenon are univariate type and, usually, use the linear regression to predict the behavior of one variable from the variation of other; among others, Simple Linear Regression method and Averages by Thermal Sensation Interval method can be identified, although it is also possible to find studies based on the ANSI/ASHRAE 55 method or on machine-learning algorithms. This document describes a procedure that allows three-stages data to be statistically processed: Database capture, Database preparation, and Data Correlation. In each case, the steps to be followed are specified and different statistical alternatives are indicated to achieve certainty in the results. From different studies specialized in thermal comfort, it is possible to identify that the Averages by Thermal Sensation Interval method offers results with greater consistency and causality regarding the perceived thermal sensation and its phenomenological correspondence with the monitoring of the environmental conditions. As a complementary resource, a personalized spreadsheet whit the three methods described in this paper is including.