Predicción de la demanda de pasajeros a clústeres de estaciones del Metropolitano usando métodos de Data Mining, la metodología Box-Jenkins y Sarima

• Autor: Roque Rojas, Edwin
• Formato: tesis de grado
• Fecha de Publicación: 2022
• Institución: Universidad de Lima
• Repositorio: ULIMA-Institucional
• Lenguaje: español
• OAI Identifier: oai:repositorio.ulima.edu.pe:20.500.12724/16675
• Enlace del recurso: https://hdl.handle.net/20.500.12724/16675
• Nivel de acceso: acceso abierto
• Materia: Transporte de pasajeros; Minería de datos; Prospectiva; Transporte urbano; Passanger transport; Data mining; Forecasting; Urban transportation; Lima Metropolitana (Perú); https://purl.org/pe-repo/ocde/ford#2.02.04

The level of passenger demand for the Metropolitan service has increased and the planning called JICA, currently used, is not enough, causing the saturation of passengers in their 38 stations. According to experts and reports made by the Metropolitan Municipality of Lima and ProTransporte in 2018, claim that the maximum station capacity of 700,000 passengers was exceeded daily, which was planned, being twice as much as 2010 and suggesting updating demand planning. So, it was proposed to predict the passenger demand of station clusters using SARIMA from a spatio-temporal analysis using two data mining methods and the Box-Jenkins methodology to get the best possible model for cluster. The results of the spatio-temporal analysis showed similar behavior between stations when grouped into clusters with weekly seasonality. The models didn't make a correct prediction for the annual holidays, as they were interpreted as outlier´s values, so the demand that recorded these dates was replaced to make the models more accurate; finally getting good results with a RMSPE, MAPE and ¿ 2 between 6.37% - 8.13%, 4.19% - 5.93% y 0.91 - 0.98 respectively between the four models, below the ceiling for each forecast metric that was proposed as targets. Despite the problem, model predictions can be used to optimize the Metropolitan's resources in the distribution of its buses, adequately taking care of the demand that saturates its stations, not counting the annual holidays.