| Since the early 1970's, disaggregate demand analysis has been widely adopted in transportation research. It provides significant improvements relative to studies based on aggregate data which prevailed in the 1960's. Although the disaggregate model focuses on individual level behavior, the estimated model parameters are fixed across individuals. To incorporate unobserved taste variations, recent developments in disaggregate choice modeling allows for the parameters to vary across individuals, represented by a random distribution. The most popular approach is the Mixed Logit model. But Mixed logit model only recognizes the differences among individuals, it does not distinguish individuals who respond differently to travel service changes.; Hierarchical Bayesian method provides a natural way to obtain individual level inferences. This research illustrates the importance of individual level analysis using studies across two research fields, transportation and marketing.; The transportation study contributes to the literature by providing a method to combine the revealed and stated preference data sources, even though the revealed preference data have limited information that prevents us from using the conventional method of jointly estimating revealed and stated preference choice models. We also conduct analysis on the distribution of value of time based on the individual level inferences, which the current literature studies VOT only through the population level inferences.; The second part of the thesis studies the individual level targeting issue in Marketing. Individual level targeting is a new trend, but costly to the firm, the question is how to quantify the benefit of individual level targeting. The conventional method can not be applied to the case when the company has already been involved in individual level targeting. For example, the pharmaceutical companies spend a lot of money and effort to collect data, and conduct analysis of each physician's behavior when they decide how to target each individual physician. In this case, the conventional method would not work. We proposed a method to allow us to quantify the benefit of individual level targeting while the firm has already engaged in targeting. It contributes to the literature by studying both the demand and supply models at individual level. |
