A blending ensemble learning model for crude oil price forecasting is proposed to efficiently capture diverse fluctuation profiles in forecasting crude oil prices. The model combines various machine learning methods, including k-nearest neighbor regression, regression trees, linear regression, ridge regression, and support vector regression. Data for Brent and WTI crude oil prices at various time series frequencies are used to validate the proposed approach. The model is benchmarked against existing individual and ensemble learning methods, such as lasso regression, bagging lasso regression, boosting, random forest, and support vector regression. The results show that the proposed blending-based model outperforms existing forecasting models in terms of forecasting errors for both short- and medium-term horizons. The study analyzes the two most commonly used crude oil data sets: Brent and West Texas Intermediate (WTI). Daily data are converted into weekly and monthly models for short- and medium-term forecasting. Statistical characteristics of the Brent and WTI time series data are analyzed, and necessary data preprocessing and mode changes are performed to enable predictions at different time series periods. The main contributions of this paper are: (1) a multiscale model based on blending ensemble learning that predicts the short-term and medium-term crude oil prices, which allows us to break the limitations of a single time series decomposition analysis. (2) The superiority of the proposed blending ensemble model is demonstrated compared to the state-of-the-art crude oil price forecasting models used as a reference. The results of the Diebold Mariano (DM) test confirm the dominance of the proposed forecasting model. The study proposes a blending ensemble machine learning model that combines five diverse predictors: k-nearest neighbor regression, linear regression, regression tree, support vector regression, and ridge regression. The model, referred to as LKDSR, is evaluated using various error measures, including MAE, MSE, RMSE, MPD, and sMAPE. The results show that the LKDSR model outperforms other models in terms of forecasting accuracy for both daily and monthly crude oil price forecasts. The DM test confirms the statistical significance of the proposed model's performance. The study also compares the performance of the LKDSR model with other models, including Lasso, SVR, AdaBoost, regression tree, LGB, CatBoost, and random forest. The results show that the LKDSR model performs well in forecasting both daily and monthly crude oil prices. The study concludes that the proposed blending ensemble learning model is effective in forecasting crude oil prices and provides more accurate predictions than existing models.A blending ensemble learning model for crude oil price forecasting is proposed to efficiently capture diverse fluctuation profiles in forecasting crude oil prices. The model combines various machine learning methods, including k-nearest neighbor regression, regression trees, linear regression, ridge regression, and support vector regression. Data for Brent and WTI crude oil prices at various time series frequencies are used to validate the proposed approach. The model is benchmarked against existing individual and ensemble learning methods, such as lasso regression, bagging lasso regression, boosting, random forest, and support vector regression. The results show that the proposed blending-based model outperforms existing forecasting models in terms of forecasting errors for both short- and medium-term horizons. The study analyzes the two most commonly used crude oil data sets: Brent and West Texas Intermediate (WTI). Daily data are converted into weekly and monthly models for short- and medium-term forecasting. Statistical characteristics of the Brent and WTI time series data are analyzed, and necessary data preprocessing and mode changes are performed to enable predictions at different time series periods. The main contributions of this paper are: (1) a multiscale model based on blending ensemble learning that predicts the short-term and medium-term crude oil prices, which allows us to break the limitations of a single time series decomposition analysis. (2) The superiority of the proposed blending ensemble model is demonstrated compared to the state-of-the-art crude oil price forecasting models used as a reference. The results of the Diebold Mariano (DM) test confirm the dominance of the proposed forecasting model. The study proposes a blending ensemble machine learning model that combines five diverse predictors: k-nearest neighbor regression, linear regression, regression tree, support vector regression, and ridge regression. The model, referred to as LKDSR, is evaluated using various error measures, including MAE, MSE, RMSE, MPD, and sMAPE. The results show that the LKDSR model outperforms other models in terms of forecasting accuracy for both daily and monthly crude oil price forecasts. The DM test confirms the statistical significance of the proposed model's performance. The study also compares the performance of the LKDSR model with other models, including Lasso, SVR, AdaBoost, regression tree, LGB, CatBoost, and random forest. The results show that the LKDSR model performs well in forecasting both daily and monthly crude oil prices. The study concludes that the proposed blending ensemble learning model is effective in forecasting crude oil prices and provides more accurate predictions than existing models.