Behavior prediction of telecom consumers´ choice with packages based on improved nested logit model

Traditional nested logit model assumes that the consumers are complete rationality when predicting the choice of consumer behavior. But in actual, consumers are often limited rationality. Reference dependent model analyzes the consumers´ behavior with limited rationality, and it extends the value function of prospect theory in multiple attributes. This paper will improve the traditional nested logit model based on reference dependence. We will consider multiple attributes and multiple reference points, and predict the choice behavior of telecom consumers with limited rationality about packages. Finally, in the example analysis, we will utilize the actual data of some telecom operator to analyze the original and improved models. Through the nested logit regression analysis using software STATA, the results will verify the effectiveness of the improved model further.