Improvements and application of normal cloud positive regression model

The normal cloud model, as an important model for simultaneously studying the randomness and fuzziness of uncertainty problems, not only covers various disciplines and fields in practical applications but has also been promoted by scholars in theoretical research since its publication. This article introduces the normal cloud model and the cloud least squares method based on discrete envelope distance. In practical applications, the cloud least squares method may generate low accuracy or results that contradict the actual situation. To solve these problems, this article presents the improvements of the normal cloud linear regression model, which leads to the normal cloud ridge regression (NCR) and normal cloud positive regression models. Their advantages have been illustrated through simulation data analysis respectively. Finally, the model was applied to actual stock data to demonstrate the value of our modification once again.