The increasing availability of individual-level longitudinal data provides the opportunity to better understand travelers'\ day-to-day learning process of their choice alternatives, which enables potentially more accurate predictions of choice patterns in a network with uncertainties. In this thesis, an instance-based learning (IBL) model for travel choice is developed within route-choice context, where on each day a traveler's decision depends on her entire choice history in the past. Learning in this model is based on the power law of forgetting and practice, which is shown to be capable of capturing various psychological effects embedded in travelers'\ day-to-day learning process, including the recency effect, hot stove effect and payoff variability effect. Estimation results based on empirical data show that the IBL model reveals higher sensitivity to perceived travel time and achieves better model fit compared to a baseline learning model. Cross-validation experiments suggest that the forecasting ability of the IBL model is consistently better than the baseline learning model. Despite the above-mentioned advantages of the IBL model, the common problem of missing initial observations in longitudinal data collection can lead to inconsistent estimates of perceived value of attributes in question, and thus inconsistent parameter estimates. In this thesis, the stated problem is addressed by treating the missing observations as latent variables. The proposed method is implemented in practice as maximum simulated likelihood (MSL) correction with two sampling methods in an instance-based learning model for travel choice, and the finite sample bias and efficiency of the estimators are investigated. Monte Carlo experimentation based on synthetic data shows that both the MSL with random sampling (MSLrs) and MSL with importance sampling (MSLis) are effective in correcting for the endogeneity problem in that the percent error and empirical coverage of the estimators are greatly improved after correction. The methods are applied to an experimental route-choice dataset to demonstrate their empirical application. Hausman-McFadden tests show that the estimators after correction are statistically equal to the estimators of the full dataset without missing observations, confirming that the proposed methods are practical and effective for addressing the stated problem. Apart from modeling travelers'\ day-to-day learning process for travel choice, day-to-day driving behavior intervention is also studied in this thesis. A study of Mitigation Techniques to Modify Driver Performance to Improve Fuel Economy, Reduce Emissions and Improve Safety was undertaken as part of the Massachusetts Department of Transportation (MassDOT) Research Program. Major conclusions include: 1) Real-time feedback has a significant effect in reducing speeding and aggressive acceleration. 2) Training has a significant effect in reducing idling rate in the first month after training. 3) Combining training and feedback is expected to significantly improve fuel economy, reduce emissions and improve safety.