(A) trace analysis of visitors for science museum

Abstract

Analyzing visitor demographics and evaluation of exhibitions are two important tasks for museums. One obstacle is that such surveys take large amount of labor and require active participation and effort from visitors, making it difficult to conduct investigation frequently. To relieve this problem, we propose an algorithm for inferring the composition of age, gender, companion, intention to revisit, and satisfaction of museum visitors from trace (location and acceleration) data collected by necklace type devices. We generated features from the trace data for our inference algorithm based on the maximum distance between ECDFs (Empirical Cumulative Distribution Function) of feature variables. For generalizability of our work, we suggested a time adjustment methodology using expected time of stay information provided by museums.