When designing the environment used by an unspecified number of people based on the evaluation of users, it is necessary to plan reflecting the variation of the evaluation criteria of a large number of people and the attributes of the people assumed at designing. In the visual environment design of the room targeting a large number of people, it is necessary to give flexibility to the design by introducing various design levels such as an rating level and an evaluation probability. The rating level is the degree of visibility of the visual target and the evaluation probability is the probability that the rating level is achieved.
In this paper, in order to construct a design method that can reflect the rating level and the evaluation probability as the design level, the method of predicting the readability rating based on the regression coefficient by logistic regression that regress the relationship between the three visual factors and the readability of documents is presented.
Based on the result of the subjective evaluation experiment of readability in the previous report, the relationship between three visual factors and readability is regressed by logistic regression, which is the linear regression model with setting the logarithmic odds of the probability being less than the neighboring boundary of readability category as the objective variable and three visual factors as the explanatory variable. With introducing a method of constructing a readability scale from the probability that can be predicted by the regression, it is proposed to formulate the relationship between three visual factors and readability.
Consistency is examined on readability of the experimental data in the previous report and that obtained by manual regression of that data and that predictted by the three kind of regression models. For the formulation of the relationship between three visual factors and the design level, the regression model is chosen as a most suitable one in which the regression coefficient is determined for each different adjacent category boundary and for each different range of adaptation luminance. The good relationship between three visual factors and readability predicted by this chosen model is shown and a useful readability prediction method for visual environment design is presented in this paper. The prediction method shown in this paper is useful for facilitating examination of visibility while introducing various design levels in visual environment design.
