Support vector machine (SVM)
Crashes are infrequent, yet road agencies must decide where to investigate or intervene before the next one occurs. That creates a prediction problem: can the conditions of a road segment help estimate its future crash count? Traffic volume, geometry, and other recorded features may interact in ways a simple straight-line rule misses.
Past road examples support two competing forecasts: a basic statistical prediction and a support-vector approach that can learn a more flexible relationship. Both models then predict roads they did not see during fitting. The important result is not that a machine-learning model explains the cause of every collision; it does not. The value is a more informative forecast for prioritizing scarce safety resources, provided performance is checked on unseen data. The video presents this as a modeling workflow, not as a claim that a particular road is safe or dangerous merely because of its score.
Large language models can generate many plausible travel choices, but plausible is not the same as representative of real travelers. In the same choice situation, a human sample and an artificial sample may favor different options. If those differences are ignored, adding more synthetic responses can make a demand model more confident about the wrong behavior.
Watch film
A road does not evolve in isolation. Speed on one link can be affected by traffic arriving from upstream, queues spilling back from downstream, and conditions on neighboring links. Forecasting several steps ahead therefore requires more than extending each road's own recent trend.
Watch film
Traffic sensors do not provide a perfect record. Some readings vanish at random, and a broken detector may leave a whole stretch of time blank. Filling those gaps matters because a missing value can distort both the picture of past congestion and the models trained to forecast it.
Watch film