Aplicación informática para la predicción de inventarios de combustible en estaciones de servicio Primax.

Abstract

Inventory control in a fuel vending company and being able to predict the demand for the product is important for its management and to be able to predict and meet the demands of customers. This project sought to implement a technological tool with a predictive algorithm to value the stock and plan immediate and future orders, for which descriptive research and the mixed approach and interviews with members of the company were used. Theories and concepts on data analytics and other topics were reviewed, in addition to the most appropriate technological tools such as Python and Prophet for the development of the algorithm, along with Sql Server and Power BI used in the gas station. From the interview it was learned that the inventory is done manually and is delayed, so implementing an application would help to improve fuel management. For the development of the tool, data from 109 gas stations and the fuels to be sold were collected and analyzed, selecting the dominant variables; the project included an ETL process to prepare and clean the data, and the implementation of Prophet to predict daily volumes in 2024, used for its ability to capture seasonality and trends. The results, along with actual values, were stored in SQL Server, and visualizations were performed in Power BI to facilitate interpretation and decision making.