Bibliographic citations
Juarez, V., (2023). Construcción de un modelo predictivo de productividad de las palas aplicando el aprendizaje automático Random Forest Regressor de Machine Learning [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/27171
Juarez, V., Construcción de un modelo predictivo de productividad de las palas aplicando el aprendizaje automático Random Forest Regressor de Machine Learning [Tesis]. PE: Universidad Nacional de Ingeniería; 2023. http://hdl.handle.net/20.500.14076/27171
@misc{renati/712989,
title = "Construcción de un modelo predictivo de productividad de las palas aplicando el aprendizaje automático Random Forest Regressor de Machine Learning",
author = "Juarez Racchumi, Victor Alexander",
publisher = "Universidad Nacional de Ingeniería",
year = "2023"
}
The Thesis focuses is the creation of a predictive model of shovel productivity with an acceptable level of confidence, which, unlike the traditional method, includes not only the variables of mined tonnage, loading time and hang time; It also considers other quantitative and qualitative variables such as queue time, spot time, loaded travel time, queue time at the destination, spot time at the destination, dumping time, empty travel time, guard, shift, month, fleet of shovel, shovel, empty efh, loaded efh, material, material type, origin phase, destination phase, hour, truck and shovel availability. The main objective is to improve the prediction of shovel productivity by applying Machine Learning, as well as to determine the best shovel productivity prediction model and determine the degree of importance of each of the independent variables for our predictor variable, through several libraries and with the help of Jupyter Notebook, ensuring that our results are the most reliable. The research will begin with the general hypothesis that the productivity model developed with Machine Learning's Random Forest Regressor improves the prediction of blade productivity than other models. For the machine learning model, all the variables present in the DataSet were used, including the qualitative variables, which were categorized before the training. Of the machine learning models, the model that best predicts the productivity of the shovels is the “Random Forest Regressor“ with an accuracy of 99.7% for training and 91.1% for the DataSet for testing; while the worst machine learning model is “Dummy Regressor“ with - 0.03% accuracy. In addition, the variable of greater degree of importance is the Shovel Fleet with 24.92%.
This item is licensed under a Creative Commons License