Assignment framework
Part of Thesis summary. How the method turns one stay into a probability for building, street and open space.
- Framework overview: the four steps, from a stay to probabilities
- Candidate selection: which nearby places are considered
- The four scores for each candidate: f_distribution, f_distance, f_temporal, f_directional
- Two extra inputs added during calibration: share_type, sigma
How the model learned its weights: Calibration.
Meetings: MNL model development (first learned weights), RQ1 rebuilt with grouped CV (one shared weight for f_directional).
Linked from Framework overview · Thesis summary · share_type · sigma