AI-Aquatica v2.0.0 uses a src/-based Python package layout and separates the software into domain-oriented layers.
src/ai_aquatica/
core/ Dataset, pipeline and result abstractions
io/ CSV, Excel, JSON, SQL, MongoDB and API loaders/importers
preprocessing/ Cleaning, missing-value handling and transformations
analysis/ Descriptive statistics and time-series analysis
modeling/ Classical ML, clustering, anomaly detection and synthetic data helpers
hydrochemistry/ Charge/ion-balance diagnostics and hydrochemical quality control
visualization/ Exploratory and model-oriented plotting
reporting/ Standalone HTML report generation and templates
datasets/ Bundled reproducible example datasets
The main SoftwareX workflow is exposed through WaterQualityPipeline, which combines dataset validation, descriptive analysis, ion-balance diagnostics, preprocessing, PCA, random-forest modelling and HTML report generation.
The former flat module layout was intentionally removed because it made the repository harder to review, maintain and describe as scientific software. The current structure is designed to support future extensions such as additional hydrochemical indices, geospatial/remote-sensing integration and richer reporting.