TPA: A Two-Phase Approach Using Simulated Annealing for the Optimization of Census Taker Routes in Mexico

Abstract EN

Censuses in Mexico are taken by the National Institute of Statistics and Geography (INEGI).

In this paper a Two-Phase Approach (TPA) to optimize the routes of INEGI’s census takers is presented.

For each pollster, in the first phase, a route is produced by means of the Simulated Annealing (SA) heuristic, which attempts to minimize the travel distance subject to particular constraints.

Whenever the route is unrealizable, it is made realizable in the second phase by constructing a visibility graph for each obstacle and applying Dijkstra’s algorithm to determine the shortest path in this graph.

A tuning methodology based on the irace package was used to determine the parameter values for TPA on a subset of 150 instances provided by INEGI.

The practical effectiveness of TPA was assessed on another subset of 1962 instances, comparing its performance with that of the in-use heuristic (INEGIH).

The results show that TPA clearly outperforms INEGIH.

The average improvement is of 47.11%.