In this work, a regression model for predicting product prices was developed using the PyTorch framework. The model building process includes several stages: data analysis and preparation, neural model architecture development, parameter selection, model training using the early stopping method to prevent overtraining and improve generalization ability. Key metrics such as mean absolute error (MAE) and mean square error (MSE) were used to evaluate the model. Additionally, two classic machine learning models, RandomForestRegressor and GradientBoostingRegressor, were built. The parameters for these models were optimized using RandomizedSearchCV, allowing the identification of the most suitable hyperparameters for each model. This approach helped improve their accuracy and robustness against overfitting. To assess the effectiveness of each model, their results were compared based on key metrics such as MAE and MSE on both training and test datasets. Additionally, graphs illustrating the relationship between actual and predicted values were generated for each model, visually assessing prediction accuracy and the nature of possible errors. As a result of the research, it was determined that the classic machine learning model, namely RandomForestRegressor, demonstrates better metrics and the highest forecasting accuracy.