Logistic Profile Generation via Clustering Analy

Abstract

The process of characterizing a city to generate logistic profiles involves the analysis of many different aspects. These profiles are based on secondary sources of data, mainly road network infrastructure, socioeconomic data and population density. Following previous research, the final profiles are given by a K-Means algorithm, which uses principal component analysis (PCA) for correlation analysis. A caveat in this method is that prior research has shown that PCA is sensitive to outliers and high dimensionality, which may mislead the following analysis and research. As such, this paper proposes a methodology to evaluate the performance of different clustering techniques to generate logistic profiles, applying it to a case study in the city of Lima, Perú.