The purpose of this study is to propose a quantitative method for consumers' preference analysis using picture drawings. In this method, the picture drawings were broken down into its component elements. The component elements are analyzed by using singular value decomposition method, and the latent decision frames are estimated. The consumers' preferences are explained by the estimated decision frames using regression analysis. We demonstrated an example of this method and conducted a survey that the participants were asked to draw a picture of their feelings towards McDonalds and Mos Burger, which are considered competing hamburger restaurants in Japan. The participants in this study were 150 university students. Based on the estimated decision frames, we predicted their preference using logistic regression analysis method. The results indicate that the decision frames projected from picture drawing data explained participants' preference, and the component elements of picture drawings affecting preference were assessed quantitatively.