Implementation Of Machine Learning and BPM To Increase Profitability in A Travel Agency

Villegas G.B.H.; Aguilar A.M.M.; Ortecho E.A. · Proceedings of the World Congress on Mechanical, Chemical, and Material Engineering (2025) · DOI: 10.11159/icmie25.128

Resumen

This research addresses the low accuracy in monthly tourist projections for a travel agency, with a Mean Absolute Percentage Error (MAPE) of 29.49%, resulting in an opportunity cost of $3.9 million in the travel package purchasing process. A solution was implemented using Machine Learning (ML) techniques, Support Vector Regression-Seasonal Autoregressive Integrated Moving Average with Exogenous Regressors (SVR-SARIMAX), and Business Process Management (BPM) to reduce the MAPE and improve the purchasing process by redesigning and adding a negotiation activity. The ML model was trained on historical data of international tourist arrivals from 2013 to 2023, and 2023 purchase records. Using Python and Arena software, the MAPE was reduced to 18.18%, with key performance indicators improving by 16%, reaching 19,907 customers, $20.7 million in revenue, and $1.8 million in opportunity cost. Furthermore, a conversion rate over 90% was achieved after implementing process improvements, demonstrating the effecti veness of ML and BPM in resource allocation and process enhancement.

Instituciones: Universidad de Lima