Soil salinization is a difficult challenge for agricultural productivity and environmental sustainability, particularly in arid and semi-arid coastal regions. This study investigates the spatial variability of soil electrical conductivity (EC) and its relationship with key cations and anions (Na⁺, K⁺, Ca²⁺, Mg²⁺, Cl⁻, CO₃²⁻, HCO₃⁻, SO₄²⁻) along the southeastern coast of the Caspian Sea in Iran. Using a combination of field-based soil sampling, laboratory analyses, and Landsat 8 spectral data, linear Multiple Linear Regression and Partial Least Squares Regression (MLR, PLSR) and nonlinear Artifician Neural Network and Support Vector Machine (ANN, SVM) modeling approaches were employed to estimate and map soil EC. Results identified Na⁺ and Cl⁻ as the primary contributors to salinity (r = 0.78 and r = 0.88, respectively), with NaCl salts dominating the region’s soil salinity dynamics. Secondary contributions from Potassium Chloride KCl and Magnesium Chloride MgCl₂ were also observed. Coastal landforms such as lagoon relicts and coastal plains exhibited the highest salinity levels, attributed to geomorphic processes and anthropogenic activities. Among the predictive models, the SVM algorithm outperformed others, achieving higher R² values and lower RMSE (RMSE_Test = 27.35 and RMSE_Train = 24.62, respectively), underscoring its effectiveness in capturing complex soil-environment interactions. This study highlights the utility of digital soil mapping (DSM) for assessing soil salinity and provides actionable insights for sustainable land management, particularly in mitigating salinity and enhancing agricultural practices in vulnerable coastal systems.

Creating a crop type map is a dominant yet complicated model to produce. This study aims to determine the best model to identify the wheat crop in the Haridwar district, Uttarakhand, India, by presenting a novel approach using machine learning techniques for time series data derived from the Sentinel-2 satellite spanned from mid-November to April. The proposed methodology combines the Normalized Difference Vegetation Index (NDVI), satellite bands like red, green, blue, and NIR, feature extraction, and classification algorithms to capture crop growth's temporal dynamics effectively. Three models, Random Forest, Convolutional Neural Networks, and Support Vector Machine, were compared to obtain the start of season (SOS). It is validated and evaluated using the performance metrics. Further, Random Forest stood out as the best model statistically and spatially for phenology parameter extraction with the least RMSE value at 19 days. CNN and Random Forest models were used to classify wheat crops by combining SOS, blue, green, red, NIR bands, and NDVI. Random Forest produces a more accurate wheat map with an accuracy of 69% and 0.5 MeanIoU. It was observed that CNN is not able to distinguish between wheat and other crops. The result revealed that incorporating the Sentinel-2 satellite data bearing a high spatial and temporal resolution with supervised machine-learning models and crop phenology metrics can empower the crop type classification process.

A method for studying the resilience of energy and socio-ecological systems is considered; it integrates approaches developed at the International Institute of Applied Systems Analysis and the Melentyev Institute of Energy Systems (MESI) of the Siberian Branch of the Russian Academy of Sciences. The article discusses in detail the methods of using intelligent information technologies, in particular semantic technologies and knowledge engineering (cognitive probabilistic modeling), which the authors propose to use in assessing the risks of natural and man-made threats to the resilience of the energy sector and social and ecological systems. More attention is paid to the study and adaptation of the integral indicator of quality of life, which makes it possible to combine these interdisciplinary studies.

Fire hazard is often mapped as a static conditional probability of fire characteristics’ occurrence. We developed a dynamic product for operational risk management to forecast the probability of occurrence of fire radiative power in the locally possible near-maximum fire intensity range. We applied standard machine learning techniques to remotely sensed data. We used a block maxima approach to sample the most extreme fire radiative power (FRP) MODIS retrievals in free-burning fuels for each fire season between 2001 and 2020 and associated weather, fuel, and topography features in northwestern south America. We used the random forest algorithm for both classification and regression, implementing the backward stepwise repression procedure. We solved the classification problem predicting the probability of occurrence of near-maximum wildfire intensity with 75% recall out-of-sample in ten annual test sets running time series cross validation, and 77% recall and 85% ROC-AUC out-of-sample in a twenty-fold cross-validation to gauge a realistic expectation of model performance in production. We solved the regression problem predicting FRP with 86% r² in-sample, but out-of-sample performance was unsatisfactory. Our model predicts well fatal and near-fatal incidents reported in Peru and Colombia out-of-sample in mountainous areas and unimodal fire regimes, the signal decays in bimodal fire regimes.

The financial services industry is experiencing a swift adoption of artificial intelligence (AI) and machine learning for a variety of applications. These technologies can be employed by both public and private sector entities to ensure adherence to regulatory requirements, monitor activities, evaluate data accuracy, and identify instances of fraudulent behavior. The utilization of artificial intelligence (AI) and machine learning (ML) has the potential to provide novel and unforeseen manifestations of interconnectivity within financial markets and institutions. This can be represented by the adoption of previously disparate data sources by diverse institutions. The researchers employed convenience sampling as the sampling method. The form was filled out over the period spanning from July 2023 to February 2024, and it was designed to be both anonymous and accessible through online and offline platforms. To assess the reliability and validity of the measurement scales and evaluate the structural model, we employed Partial Least Squares (PLS) for model validation. Specifically, we have used the software package Smart-PLS 3 with a bootstrapping of 5000 samples to estimate the significance of the parameters. The results indicate a positive and direct connection between artificial intelligence (AI) and either financial services or financial institutions. On the contrary, machine learning (ML) exhibits a strong and positive association among financial services and financial institutions. Similarly, there exists a positive and direct connection between AI and investors, as well as between ML and investors.

The goal of this work was to create and assess machine-learning models for estimating the risk of budget overruns in developed projects. Finding the best model for risk forecasting required evaluating the performance of several models. Using a dataset of 177 projects took into account variables like environmental risks employee skill level safety incidents and project complexity. In our experiments, we analyzed the application of different machine learning models to analyze the risk for the management decision policies of developed organizations. The performance of the chosen model Neural Network (MLP) was improved after applying the tuning process which increased the Test R² from −0.37686 before tuning to 0.195637 after tuning. The Support Vector Machine (SVM), Ridge Regression, Lasso Regression, and Random Forest (Tuned) models did not improve, as seen when Test R² is compared to the experiments. No changes in Test R²’s were observed on GBM and XGBoost, which retained same Test R² across different tuning attempts. Stacking Regressor was used only during the hyperparameter tuning phase and brought a Test R² of 0. 022219.Decision Tree was again the worst model among all throughout the experiments, with no signs of improvement in its Test R²; it was −1.4669 for Decision Tree in all experiments arranged on the basis of Gender. These results indicate that although, models such as the Neural Network (MLP) sees improvements due to hyperparameter tuning, there are minimal improvements for most models. This works does highlight some of the weaknesses in specific types of models, as well as identifies areas where additional work can be expected to deliver incremental benefits to the structured applied process of risk assessment in organizational policies.