ISSN :2582-9793
  • Subscribe

    • Predicting financial habits and saving propensity using machine learning and agile methodologies

    Original Research (Published On: 08-Nov-2026 )
    DOI : https://doi.org/10.54364/AAIML.2026.66356

    Norberto Ulises Roman Concha, Jhonathan Centti, José Piedra, Ronald Melgarejo, Carlos Chávez-Herrera, Luzmila Pró, Pablo Romero, Deisi Pomajulca, María Elena Ruiz and Augusto Cortez

    Adv. Artif. Intell. Mach. Learn., - (-):-

    1. Norberto Ulises Roman Concha: National University of San Marcos

    2. Jhonathan Centti: Graduated from the Professional School of Software Engineering, UNMSM, Lima-Perú

    3. José Piedra: Graduated from the Professional School of Software Engineering, UNMSM, Lima-Perú

    4. Ronald Melgarejo: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima-Perú

    5. Carlos Chávez-Herrera: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima-Perú

    6. Luzmila Pró: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima-Perú

    7. Pablo Romero: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima-Perú

    8. Deisi Pomajulca: Professors at the Faculty of Sciences at the National University of Education, UNE, Lima-Perú

    9. María Elena Ruiz: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima-Perú

    10. Augusto Cortez: Professors at the Faculty of Systems Engineering and Computer Science, UNMSM, Lima- Perú

    Download PDF Here

    DOI: 10.54364/AAIML.2026.66356

    Article History: Received on: 20-Jun-26, Accepted on: 01-Nov-26, Published on: 08-Nov-26

    Corresponding Author: Norberto Ulises Roman Concha

    Email: nromanc@unmsm.edu.pe

    Citation: Ulises Roman-Concha et al. Predicting Financial Habits and Saving Propensity Using Machine Learning and Agile Methodologies. Advances in Artificial Intelligence and Machine Learning. 2026. [Ahead of Print]. https://dx.doi.org/10.54364/AAIML.2026.66356


    Abstract

    Many economically active individuals struggle to manage their personal finances due to a lack of personalized technological tools that enable them to anticipate financial behavior. This article addresses this issue by presenting a machine learning model to predict financial habits and saving propensity. A hybrid methodology was used that integrates CRISP-DM (Cross-Industry Standard Process for Data Mining) to structure the data science process and a framework for agile software development (SCRUM) for agile project management. Heuristic rules were defined to create two variables: “Financial Habits Level” (multi-class, 1–5) and “Savings Propensity” (binary), based on sociodemographic and financial data. Seven supervised classification algorithms were compared, and XGBoost was selected as the best algorithm, as it uses advanced techniques such as SMOTE (Synthetic Minority Over-Sampling Technique) and a machine learning hyperparameter optimization technique to find good combinations of a model’s hyperparameters by randomly testing different configurations and evaluating them via cross-validation (RandomizedSearchCV), and then rigorously evaluated using an independent test set. The “Financial Habits Level” model achieved an accuracy of 80%, while the “Savings Propensity” model achieved an accuracy of 94.5% and a ROC-AUC (Receiver Operating Characteristic - Area Under the Curve) of 0.9920, demonstrating high predictive power. 


    Statistics

    Article Views: 28
    PDF Downloads: 0