This research explores the use of machine learning (ML) techniques to predict electricity consumption. It focuses on predicting the electricity demand in Puno, Peru, using a dataset with over 4 million records from ElectroPuno, the electricity distribution company. The study examines various machine learning models, including Gradient Boosting, Decision Trees, Random Forest, K-Nearest Neighbors (KNN), Linear Regression, LightGBM, XGBoost, and Neural Networks, to identify the most effective models for predicting electricity consumption. The results show that Gradient Boosting and Decision Trees are the most accurate models, achieving high R2 values (0.9113 and 0.9048, respectively) and low error metrics, indicating their ability to effectively capture consumption patterns. Random Forest also performed well, although slightly less accurately than Gradient Boosting. In contrast, models such as KNN, Linear Regression, and advanced models like LightGBM, XGBoost, and Neural Networks demonstrated lower performance, with negative R2 values and high error rates, suggesting they were either overfitted or inadequately configured for this particular dataset. The paper highlights the importance of choosing the right machine learning model based on data characteristics and complexity, recommending further exploration of hyperparameter optimization and preprocessing techniques to improve the performance of less effective models. This research contributes to better energy management and sustainable use of electricity by providing more reliable consumption predictions.