A performance and interpretability assessment of machine learning models for rainfall prediction in the Republic of Ireland DOI Creative Commons

Menatallah Abdel Azeem,

Soumyabrata Dev

Decision Analytics Journal, Год журнала: 2024, Номер 12, С. 100515 - 100515

Опубликована: Авг. 24, 2024

Rainfall prediction significantly impacts agriculture, water reserves, and preparations for flooding conditions. This research examines the performance interpretability of machine learning (ML) models rainfall in Republic Ireland. The study uses a brute force approach Leave One Feature Out (LOFO) methodology to evaluate model under highly correlated variables. Results reveal consistent across ML algorithms, with average Area Under Curve Precision-Recall (AUC-PR) scores ranging from 0.987 1.000, certain features such as atmospheric pressure soil moisture deficits demonstrating significant influence on outcomes.SHapley Additive exPlanations (SHAP) values provide insights into feature importance, reaffirming significance prediction. underscores importance selection enhancing accuracy usability

Язык: Английский

Enhancing hydrological predictions: optimised decision tree modelling for improved monthly inflow forecasting DOI Creative Commons

Osama A. Abozweita,

Ali Najah Ahmed, Lariyah Mohd Sidek

и другие.

Journal of Hydroinformatics, Год журнала: 2024, Номер unknown

Опубликована: Сен. 26, 2024

ABSTRACT The utilisation of modelling tools in hydrology has been effective predicting future floods by analysing historical rainfall and inflow data, due to the association between climate change flood frequency. This study utilised a dataset monthly for Terengganu River Malaysia, it is renowned its hydrological patterns that exhibit high level unpredictability. evaluation predictive precision effectiveness Optimised Decision Tree ODT model, along with RF GBT models, this involved several indicators. These indicators included correlation coefficient, mean absolute error, percentage relative root square Nash-Sutcliffe efficiency, accuracy rate. research results indicated models performed better than model inflows. as well showed validation average accuracies 94%, 91%, 92%, respectively. R² values were 90.2%, 84.8%, 96.0%, respectively, NES ranged from 0.92 0.94. have greater implications, extending beyond forecasting rates encompass other hydro-meteorological variables depend exclusively on input data.

Язык: Английский

Процитировано

3

Neural Networks and Fuzzy Logic-Based Approaches for Precipitation Estimation: A Systematic Review DOI Creative Commons
Andrés Felipe Ruiz-Hurtado, Viviana Vargas-Franco, Luis Octavio González Salcedo

и другие.

Ingeniería e Investigación, Год журнала: 2025, Номер 44(3), С. e108609 - e108609

Опубликована: Янв. 31, 2025

Precipitation estimation at the river basin level is essential for watershed management, analysis of extreme events and weather climate dynamics, hydrologic modeling. In recent years, new approaches tools such as artificial intelligence techniques have been used precipitation estimation, offering advantages over traditional methods. Two major paradigms are neural networks fuzzy logic systems, which can be in a wide variety configurations, including hybrid modular models. This work presents literature review on metaheuristic models based signal processes, focusing applications these estimation. The selection comparison criteria were model type, input output variables, performance metrics, fields application. An increase number this type studies was identified, mainly involving network models, tend to get more sophisticated according availability quality training data. On other hand, hybridize with There still challenges related prediction spatial temporal resolution micro-basin levels, but, overall, very promising analysis.

Язык: Английский

Процитировано

0

Design of an enhanced fuzzy neural network-based high-dimensional information decision-making model for supply chain management in intelligent warehouses DOI

Fangyuan Tian,

D. H. Yuan

Kybernetes, Год журнала: 2025, Номер unknown

Опубликована: Апрель 22, 2025

Purpose This paper aims to optimize supply chain information decision-making systems better manage complex, high-dimensional and uncertain through the integration of fuzzy logic neural network technology. Design/methodology/approach A framework based on reasoning is developed address empirical issues in traditional systems. Subsequently, an innovative radial basis function-dynamic (RBF-DFNN) model constructed, enhancing system’s capability interpret information. retains advantages dynamic networks (DFNN) while introducing anti-fuzzy layer optimizing membership function T-paradigm layers. Findings The RBF-DFNN leads creation a for chains. Experimental results indicate that this effectively utilizes K-medoids clustering algorithm accurately capture characteristics intrinsic correlations data. Parameter optimization significantly improves model’s performance, with root mean squared error (RMSE) absolute (MAE) enhanced, resulting coefficients determination rising from 95.6 97.8–99.1% compared STPF-AIMM ANFIS networks. Originality/value study contributes advancement management by developing highly intelligent refined model, intelligence level storage promoting more sophisticated operations.

Язык: Английский

Процитировано

0

A novel approach for precipitation modeling using artificial intelligence-based ensemble models DOI Creative Commons
Jazuli Abdullahi, Imran Rufai,

Nanna Nanven Rimtip

и другие.

Desalination and Water Treatment, Год журнала: 2024, Номер 317, С. 100188 - 100188

Опубликована: Янв. 1, 2024

Precipitation is the primary component of hydrologic water cycle and its accurate prediction plays a significant role for planning, management design hydraulic structures. The objective study intended to explore new approach increase efficiency precipitation in arid, semiarid humid zones. was implemented using relevant data from stations Iraq Nigeria. Support vector regression (SVR), artificial neural network (ANN) adaptive neuro-fuzzy inference system models (ANFIS) were applied single modeling 30 years monthly average data. Thereafter, nonlinear ensemble 2 linear techniques improve reliability models. Error measures as well goodness fit measure criteria employed assess performance potential Based on results this work it found that could accuracy much 38% validation phase. It also confirmed intelligence-based can efficiently be improved by application

Язык: Английский

Процитировано

2

Monthly Rainfall Forecasting Using High Order Singh’s Fuzzy Time Series Based on Interval Ratio Methods: Case Study Semarang City, Indonesia DOI Open Access

Erikha Feriyanto,

Farikhin,

Nikken Prima Puspita

и другие.

Asian Journal of Probability and Statistics, Год журнала: 2024, Номер 26(8), С. 71 - 88

Опубликована: Июль 23, 2024

Aims: Sample: To determine the effectiveness of proposed forecasting method, namely Singh's fuzzy time series based on high order (third order) interval ratios. And find out results in January 2022. Study Design: Modification Place and Duration Study: monthly rainfall data for Semarang City from 2017 to December 2021. Methodology: The method by researcher is Singh This research uses a combination Chen series. Applying Chen's section determining universe discourse () fuzzification which includes discourse, partitions, forming Fuzzy Logical Relationships Relationship Groups. Then apply part. Finally, calculate Average Forecasting Error Rate (AFER) test performance. In part, it obtained through heuristic approach building rules obtain better have an effect very small AFER values. step partition, this ratio aims reflect variations historical data. Conclusion: Based calculation value, third 0.2422%. It can be said that ratios 2021 good. forecast 2022 196.80 mm3 or into category heavy rain.

Язык: Английский

Процитировано

0

A performance and interpretability assessment of machine learning models for rainfall prediction in the Republic of Ireland DOI Creative Commons

Menatallah Abdel Azeem,

Soumyabrata Dev

Decision Analytics Journal, Год журнала: 2024, Номер 12, С. 100515 - 100515

Опубликована: Авг. 24, 2024

Rainfall prediction significantly impacts agriculture, water reserves, and preparations for flooding conditions. This research examines the performance interpretability of machine learning (ML) models rainfall in Republic Ireland. The study uses a brute force approach Leave One Feature Out (LOFO) methodology to evaluate model under highly correlated variables. Results reveal consistent across ML algorithms, with average Area Under Curve Precision-Recall (AUC-PR) scores ranging from 0.987 1.000, certain features such as atmospheric pressure soil moisture deficits demonstrating significant influence on outcomes.SHapley Additive exPlanations (SHAP) values provide insights into feature importance, reaffirming significance prediction. underscores importance selection enhancing accuracy usability

Язык: Английский

Процитировано

0