Considering the association between commuters' activity schedule and trip schedule within a bi-modal transportation system consisting of transit and auto mode, we replace the utility function in the Logit model by quantitative net utility function, and then analyze the commuters' travel mode choice problem in the morning peak with constant and linear marginal activity utility functions. It is shown that the trip-based approach always estimates the number of commuters between the constant marginal activity utility function and the linear marginal activity utility function; the number of commuters under marginal activity utility function is more sensitive than that under trip-based approach while increasing the value of unit time for early arrival. However, the number of commuters is less sensitive for both marginal activity utility function and trip-based approach. In addition, within the selected parameter range, total system cost function is a piecewise function, and find the optimal departure interval.