問題文
A logistics company stores five years of completed deliveries. Each row carries the route, the vehicle, the weather at departure, and the minutes the delivery actually took. The team wants a model that estimates the minutes for a delivery that has not happened yet. Which category of machine learning does this work belong to, and why?
選択肢
- Supervised learning, because every historical row already carries the value the model is asked to estimate, so the model can be fitted against known answers.
- Reinforcement learning, because the company wants to improve future deliveries on each route and the model will be judged by the outcome of the deliveries it influences.
- None of these, because the target is a duration in minutes rather than a category, and durations are handled by statistical summaries rather than by machine learning.
- Unsupervised learning, because the model has to find structure in the routes and the weather before the minutes column can be used, so the fitting proceeds without a target.