Integrating partial least square structural equation modelling and machine learning for causal exploration of...

Area of Science:

• Environmental Science
• Ecological Modeling
• Water Quality Management

Background:

• Understanding environmental phenomena requires robust causal analysis.
• Partial Least Squares Structural Equation Modelling (PLS-SEM) is used in ecological environment studies (EES) but often misses nonlinearities and machine learning integration.
• Harmful Algal Blooms (HABs) pose significant ecological and economic challenges.

Purpose of the Study:

• To develop and validate a novel framework for analyzing causal relationships in environmental phenomena, specifically Spring Harmful Algal Blooms (Spring HABs).
• To integrate machine learning techniques with PLS-SEM to capture nonlinearities and interactions among environmental factors.
• To identify key drivers of Spring HABs in Gaoyang Lake, Three Gorges Reservoir Region.

Main Methods:

• A hybrid framework combining PLS-SEM, Bayesian Networks (BN), Multivariate Adaptive Regression Splines (MARS), and Polynomial Regression (PR).
• BN was used to optimize causal structure for PLS-SEM.
• MARS and PR were employed to detect interactions and nonlinearities among predictors.

Main Results:

• The BN-optimized PLS-SEM structure improved the Bayesian Information Criterion (BIC) score.
• Polynomial Regression identified significant nonlinearities, leading to an optimal model with R² of 0.421 and Q²predict of 0.177.
• Euphotic depth (interacting with epilimnion depth), surface nutrient levels (total phosphorus), and meteorological factors (temperature, sun hours) were the primary drivers of Spring HABs.

Conclusions:

• The proposed framework enhances causal understanding of environmental phenomena by integrating advanced statistical and machine learning methods.
• This approach provides a more accurate model for predicting Spring HABs, accounting for complex environmental interactions.
• The findings offer a scientific basis for improved environmental management strategies to mitigate HABs.