Payment behavior model for students at a private university in Peru

Henry O. Villarreal Torres; Julio C. Angeles Morales; William J. Marín Rodríguez; Daniel Cristobal Andrade Girón; Eduardo A. Carreño

Ayuda

Payment behavior model for students at a private university in Peru

Villarreal Torres, Henry ^[1] ; Ángeles Morales, Julio ^[1] ; Marín Rodriguez, William Joel ^[2] ; Andrade Girón, Daniel ^[2] ; Carreño Cisneros, Edgardo ^[2]
1. [1] Universidad San Pedro
  
  Universidad San Pedro
  
  Chimbote, Perú
2. [2] Universidad Nacional José Faustino Sánchez Carrión
  
  Universidad Nacional José Faustino Sánchez Carrión
  
  Huacho, Perú
Localización: Salud, Ciencia y Tecnología - Serie de Conferencias, ISSN-e 2953-4860, Vol. 2, Nº. 0, 2023 (Ejemplar dedicado a: Salud, Ciencia y Tecnología - Serie de Conferencias)
Idioma: inglés
Enlaces
- Texto completo (pdf)
Resumen
- With the enactment of Law No. 29947, “Law for the Protection of the Family Economy”, students use the educational service with the payment of tuition, demonstrating a poor payment culture, failing to pay tuition until the beginning of the next semester. This motivates the university to present a state of illiquidity. The objective of the research was to develop a payment behavior classification model for students of a private university in Peru, with the purpose of predicting delinquency and compliance with payment commitments, through the implementation of strategies to improve the quality of the economic collection process. The methodology presents a research component of technological type, of propositional level, incremental innovation, the data collection was of retrospective type; with a synchronous temporal scope, because it was carried out in a short period of time, less than a year, the study population consisted of 8495 enrolled undergraduate students. The results show a classification model to predict payment behavior, using the H2O.ai platform and the R programming language, the data were obtained from computer systems, using the CRISP-DM methodology used in data science solutions. The datasets for training, validation and testing correspond to 70 %, 15 % and 15 %; obtaining the GBM Grid classification model whose performance metrics are AUC of 0,6272, AUCPR of 0,8751 and logLoss equivalent to 0,4577.