Ensemble Machine Learning of Gradient Boosting (XGBoost, LightGBM, CatBoost) and Attention-Based CNN-LSTM for Harmful Algal Blooms Forecasting

Toxins, Journal Year: 2023, Volume and Issue: 15(10), P. 608 - 608

Published: Oct. 10, 2023

Harmful algal blooms (HABs) are a serious threat to ecosystems and human health. The accurate prediction of HABs is crucial for their proactive preparation management. While mechanism-based numerical modeling, such as the Environmental Fluid Dynamics Code (EFDC), has been widely used in past, recent development machine learning technology with data-based processing capabilities opened up new possibilities prediction. In this study, we developed evaluated two types learning-based models prediction: Gradient Boosting (XGBoost, LightGBM, CatBoost) attention-based CNN-LSTM models. We Bayesian optimization techniques hyperparameter tuning, applied bagging stacking ensemble obtain final results. result was derived by applying optimal techniques, applicability evaluated. When predicting an technique, it judged that overall performance can be improved complementing advantages each model averaging errors overfitting individual Our study highlights potential emphasizes need incorporate latest into important field.

Language: Английский

---

How accurate are the machine learning models in improving monthly rainfall prediction in hyper arid environment?

Journal of Hydrology, Journal Year: 2024, Volume and Issue: 633, P. 131040 - 131040

Published: March 11, 2024

Language: Английский

---

A Novel Double Machine Learning Strategy for Producing High‐Precision Multi‐Source Merging Precipitation Estimates Over the Tibetan Plateau

Water Resources Research, Journal Year: 2024, Volume and Issue: 60(4)

Published: April 1, 2024

Abstract Precipitation estimation over the Tibetan Plateau is a critical but challenging task due to sparse gauges and high altitudes. Traditional statistic methods are often insufficient characterize nonlinear relationship between different precipitation information, while machine learning techniques, particularly deep algorithms, offer novel powerful approach improve merging accuracy of multi‐source data by efficiently capturing their spatiotemporal dynamics features. This study introduced strategy called Double Machine Learning (DML), which integrates meteorological satellite retrievals, reanalysis produce high‐precision product at 0.1° × 0.1°, daily resolution for Plateau. The quantitative evaluation DML was accomplished using both auto‐meteorological independent observations. Statistical scores indicate that new DML‐based apparently outperforms three widely‐used datasets (IMERG‐Final, GSMaP‐Gauge ERA5) proposed effectively advantages traditional learning, significantly enhancing algorithmic robustness accuracy, medium‐high rain rates in summer. Furthermore, contributions inputs final effect systematically analyzed. It found as an auxiliary variable DML, plays crucial role identifying rainy events adjusting bias estimates, especially those ungauged regions. affirms call improving estimates combining approaches. reported here recommended hydrometeorological users science community.

Language: Английский

---

Harnessing AI for solar energy: Emergence of transformer models

Applied Energy, Journal Year: 2024, Volume and Issue: 369, P. 123541 - 123541

Published: June 1, 2024

Language: Английский

---

A new regional reference evapotranspiration model based on quantile approximation of meteorological variables

Agricultural Water Management, Journal Year: 2025, Volume and Issue: 308, P. 109299 - 109299

Published: Jan. 13, 2025

---

Evaluation of Solar Radiation Prediction Models Using AI: A Performance Comparison in the High-Potential Region of Konya, Türkiye

Atmosphere, Journal Year: 2025, Volume and Issue: 16(4), P. 398 - 398

Published: March 30, 2025

Solar radiation is one of the most abundant energy sources in world and a crucial parameter that must be researched developed for sustainable projects future generations. This study evaluates performance different machine learning methods solar prediction Konya, Turkey, region with high potential. The analysis based on hydro-meteorological data collected from NASA/POWER, covering period 1 January 1984 to 31 December 2022. compares Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), Gated Recurrent Unit (GRU), GRU (Bi-GRU), LSBoost, XGBoost, Bagging, Random Forest (RF), General Regression Neural Network (GRNN), Support Vector Machines (SVM), Artificial Networks (MLANN, RBANN). variables used include temperature, relative humidity, precipitation, wind speed, while target variable radiation. dataset was divided into 75% training 25% testing. Performance evaluations were conducted using Mean Absolute Error (MAE), Root Square (RMSE), coefficient determination (R2). results indicate Bi-LSTM models performed best test phase, demonstrating superiority deep learning-based approaches prediction.

Language: Английский

---

A Comparison of Machine Learning Models for Predicting Rainfall in Urban Metropolitan Cities

Sustainability, Journal Year: 2023, Volume and Issue: 15(18), P. 13724 - 13724

Published: Sept. 14, 2023

Current research studies offer an investigation of machine learning methods used for forecasting rainfall in urban metropolitan cities. Time series data, distinguished by their temporal complexities, are exploited using a unique data segmentation approach, providing discrete training, validation, and testing sets. Two models created: Model-1, which is based on daily Model-2, weekly data. A variety performance criteria to rigorously analyze these models. CatBoost, XGBoost, Lasso, Ridge, Linear Regression, LGBM among the algorithms under consideration. This study provides insights into predictive abilities, revealing significant trends across phases. The results show that ensemble-based algorithms, particularly CatBoost outperform both emerged as model choice throughout all assessment stages, including testing. MAE was 0.00077, RMSE 0.0010, RMSPE 0.49, R2 0.99, confirming CatBoost’s unrivaled ability identify deep intricacies within patterns. Both had indicating remarkable predict trends. Significant XGBoost included 0.02 0.10, handle longer time intervals. Regression varies. Scatter plots demonstrate robustness demonstrating capacity sustain consistently low prediction errors dataset. emphasizes potential transform meteorology planning, improve decision-making through precise forecasts, contribute disaster preparedness measures.

Language: Английский

---

Ultra-short-term wind power prediction method combining financial technology feature engineering and XGBoost algorithm

Heliyon, Journal Year: 2023, Volume and Issue: 9(6), P. e16938 - e16938

Published: June 1, 2023

The input features of existing wind power time-series data prediction models are difficult to indicate the potential relationships between data, and methods based on deep learning, which makes convergence slow be applied actual production environment. To solve above problems, an ultra-short-term model XGBoost algorithm combined with financial technical index feature engineering variational ant colony is proposed. innovatively applies indicators from time series creates a class that can highly condense data. A bionic used search for best computational parameters reduce reliance experts' experience. Taking German company Tennet set as example, proposed in this study has mean absolute error 0.859 root square 1.329, it takes only 244 ms complete prediction. Thus, provides new solution

Language: Английский

---

A novel kernel ridge grey system model with generalized Morlet wavelet and its application in forecasting natural gas production and consumption

Energy, Journal Year: 2023, Volume and Issue: 287, P. 129630 - 129630

Published: Nov. 10, 2023

Language: Английский

---

Boosting Algorithm to Handle Unbalanced Classification of PM_2.5 Concentration Levels by Observing Meteorological Parameters in Jakarta-Indonesia Using AdaBoost, XGBoost, CatBoost, and LightGBM

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 35680 - 35696

Published: Jan. 1, 2023

Air quality conditions are now more severe in the Jakarta area that is among world's top eight worst cities according to 2022 Quality Index (AQI) report. In particular, data from Meteorological, Climatological, and Geophysical Agency (BMKG) of Republic Indonesia, latest outcomes air surrounding areas, says PM2.5 concentrations have increased peaked at 148μ g/m3 2022. While a classification system for this pollution necessary critical, observation measured through BMKG Kemayoran station, Jakarta, turns out be identified as an unbalanced class. Thus, work, we perform boosting algorithm supervised learning handle such toward concentration levels by observing meteorological patterns during 1 January 2015 7 July The algorithms considered research include Adaptive Boosting (AdaBoost), Extreme Gradient (XGBoost), Categorical (CatBoost), Light Machine (LightGBM). Our simulations proven can significantly reduce bias combination with variance reduction within-class coefficients, class values: good 62%, moderate 34%, unhealthy 59%, respectively.

Language: Английский

---