Statistical estimation of integrated travel demand models (TDMs) is tedious and involves an iterative process which requires expert knowledge and time to perform several tasks including but not limited to: preprocessing data, selecting appropriate attributes, selecting appropriate models, finding model parameters, and critically analyzing results (Castiglione et al., 2015). Model estimation involves varying some free parameters, hyperparameters, to produce several models which are compared to select the most superior one. Considerable time and expertise are required to tune hyperparameters and consequently select appropriate models; which is prone to subjective error. The subjectivity of tuning hyperparameters for integrated TDMs makes published results difficult to reproduce, extend, and development is more an art than a science (Bergstra et al., 2011). There is a need for methods to minimize subjectivity in hyperparameter tuning and model selection.