The traditional transportation forecasting process, formally known as the Urban Transportation Modeling System (UTMS), utilizes a sequential four-step mathematical model:

The Lowry Model explicitly links transportation and land use by assuming that basic employment (industries that export goods outside the region) drives population growth, which in turn drives retail/service employment. It uses gravity model concepts to distribute these populations and service jobs across zones based on travel times, serving as the foundational input for the Trip Generation step.

Transportation infrastructure directly impacts accessibility. When a new highway is built, land becomes more accessible, which spurs new development (e.g., suburbs and shopping centers). This new development, in turn, generates more trips, eventually congesting the new highway. Planning models must account for this continuous, cyclical feedback loop between land use changes and transportation investments.

ABMs are inherently microscopic and rely heavily on Utility Maximization Theory applied across daily schedules. Instead of calculating isolated origin-destination pairs, the model generates a daily tour (a sequence of trips starting and ending at home). The utility $U_{it}$ of an individual $i$ choosing a daily activity pattern $t$ is formulated as:

Where $V_{it}$ is the systematic utility (dependent on the person's income, age, and the travel time/cost of the entire tour) and $\varepsilon_{it}$ represents unobserved random factors. ABMs use Monte Carlo simulation to predict the exact timing, mode, and destination of every stop on the tour for millions of synthetic individuals, providing a far more accurate forecast of complex modern congestion patterns (like peak spreading) than the 4-step model.

The most common discrete choice formulation. It assumes that the unobserved random components of utility ($\varepsilon$) are independent and identically distributed (Gumbel distribution). The probability ($P$) of choosing mode $m$ is:

Where $V_m$ is the deterministic utility function:

A critical flaw of the standard MNL model is the Independence of Irrelevant Alternatives (IIA) property (e.g., the red bus/blue bus paradox). If a new transit option is introduced, MNL assumes it will draw riders proportionally from all other modes, including driving.

The Nested Logit Model solves this by grouping similar choices. For example, the top-level nest might be a choice between "Auto" and "Transit." The bottom-level nest under "Transit" would be the choice between "Bus" and "Rail." This correctly models that a new rail line will primarily cannibalize bus ridership rather than pulling drivers out of their cars.

The primary tool for evaluation is Cost-Benefit Analysis. Alternatives with a Benefit-Cost Ratio (BCR) greater than 1.0 are generally considered economically viable. However, planners must also perform Environmental Impact Assessments (EIAs) to account for non-monetary impacts like noise pollution and habitat disruption.