Resumen

CASTILLO MENDEZ, ROBINSON  y  CAMACHO CASTRO, JULIÁN ANDRÉS. Frost and relevant meteorological variables forecast in agriculture in the Sabana de Bogotá using machine learning. Ing. Desarro. [online]. 2025, vol.43, n.1, pp.122-139.  Epub 03-Ene-2025. ISSN 0122-3461.  https://doi.org/10.14482/inde.43.01.155.454.

Taking into account historical information on climatological and frost variables, it is possible to improve decisions made in agricultural activities, seeking to determine patterns that guarantee greater yield and quality of crops and implementing forecast models based on machine learning (ML). This work presents the development of a ML model that allows determining the behavior of the meteorological variables, temperature, rainfall, and relative humidity, as well as frost, in the Sabana de Bogotá. The starting point was the creation of a historical database of these variables from 2010 to April 2023, considering information from ten different meteorological stations in the region. It has been necessary to implement data imputation techniques in information gaps. To determine the model with the response closest to reality, a model based on multiple linear regression and another on artificial neural networks were developed. According to the results and the level of absolute error, the second model approximates its forecasts closer to the real data. The work developed can be an essential tool to generate an early warning system that helps farmers in the Sabana de Bogotá.

Palabras clave : feedforward neural network; frost forecasting; machine learning; meteorological conditions; multiple linear regression.

        · resumen en Español     · texto en Español     · Español ( pdf )