Differentiating Population Spatial Behavior Using Representative Features of Geospatial Mobility (ReFGeM)

Abstract

Understanding how humans use and consume space by comparing stratified groups, either through observation or controlled study, is key to designing better spaces, cities, and policies. GPS data traces provide detailed movement patterns of individuals but can be difficult to interpret due to the scale and scope of the data collected. For actionable insights, GPS traces are usually reduced to one or more features which express the spatial phenomenon of interest. However, it is not always clear which spatial features should be employed, and substantial effort can be invested into designing features which may or may not provide insight. In this paper we present an alternative approach: a standardized feature set with actionable interpretations that can be efficiently run against many datasets. We show that these features can distinguish between disparate human mobility patterns, although no single feature can distinguish them alone.

La dimensión fractal como parte de un modelo computacional para predecir el espesor de películas delgadas de ZnO

“…La dimensión Hausdorff Besicovitch o dimensión fractal de un conjunto A restringido en ℝ 𝑛 , es un número real usado para caracterizar la complejidad geométrica de A; en la cual el conjunto A de denomina un fractal si su dimensión Hausdorff Besicovitches estrictamente mayor que su dimensión topológica (Chaudhuri & Sarkar, 1995). La dimensión fractal es una expresión lógica de la complejidad espacial que mide la complejidad espacial y codifica el número (a menudo fraccionario) de grados de libertad necesarios para describir una curva o conjunto de datos en particular (Zhang et al, 2019b). La dimensión fractal, como caracterización cuantitativa y parámetro básico del fractal, es un principio importante de la teoría fractal (Bi et al, 2015).…”

ALF-Score+ - Personalization of a Predictive Network-Based Walkability Scoring System

“…People tend to have many similarities in their daily routines such as preferences, profession or hobbies [12][13][14] [15]. For instance, many people work regular shifts between 9am and 5pm with most taking their lunch break at or around noon.…”

A multi‐agent system for itinerary suggestion in smart environments

“…The authors of [8] studied the discriminative power of several features in distinguishing between different datasets of human mobility traces. The features considered include as follows: the convex hull enclosing all users positions, the entropy of individual trajectories and the degrees of freedom required to describe a particular trajectory.…”

“…Nevertheless, supporting the itinerary decisions of unknown users is especially important in scenarios where users have only few interactions with the RS, for instance, the visitors of a museum. An emerging way [7][8][9] to deal with this challenging problem lies in adopting a reinforcement learning strategy, considering the two-faced aim of a RS: on the one hand, it aims to maximise user satisfaction with the immediate itinerary suggestion (exploitation of known knowledge), on the other hand, it wants to gain new knowledge about user preferences (exploration of new information) to improve long-term reward, although with some effect on short-term satisfaction. This dilemma is typically conjugated in the context of RSs as a Multi-Armed Bandit problem [7,10].…”