Fitting Analysis of Subjective Evaluation Data Based on Unity Audio-visual Interaction

In this paper, the subjective evaluation data of image localization in Unity audio-visual interaction platform is analyzed. Three classical models are used to fit the experimental data. The regression parameters were compared under three classical models by curve fitting. Regression analysis shows that the linear model has the best fitting effect under the confidence interval of 95%. This paper further analyzes the residual processing, integrates the linear model with the exponential model, and the improved model can better retain effective data, and the regression effect is better than the classical data model.