geosmith 0.1.1

*Smith for ML models

GeoSmith

*Smith for geospatial ML models with strict 4-layer architecture.

Features

• 4-Layer Architecture: Clean separation between objects, primitives, tasks, and workflows
• Optional Integrations: Works seamlessly with PlotSmith, AnomSmith, and TimeSmith
• Geostatistics: Variogram analysis, kriging, IDW interpolation
• Mining Tools: Block models, drillhole processing, ore grade estimation
• I/O Support: GRDECL, LAS, vector/raster formats
• No Hard Dependencies: Core works with just NumPy and Pandas

Installation

# Core installation
pip install geosmith

# With optional integrations
pip install geosmith[plotsmith,anomsmith,timesmith]

# Everything
pip install geosmith[all]

Quick Start

from geosmith import PointSet, OrdinaryKriging
from geosmith.primitives.variogram import fit_variogram_model

# Create sample points
points = PointSet(coordinates=sample_coords)

# Fit variogram and kriging
variogram = fit_variogram_model(lags, semi_vars)
kriging = OrdinaryKriging(variogram_model=variogram)
kriging.fit(points, values)

# Predict
result = kriging.predict(query_points)

*Smith Family Integration

GeoSmith works seamlessly with other *Smith libraries:

• PlotSmith: Publication-ready plots (optional) - https://github.com/kylejones200/plotsmith
• AnomSmith: Spatial anomaly detection (optional) - https://github.com/kylejones200/anomsmith
• TimeSmith: Time series compatibility (optional) - https://github.com/kylejones200/timesmith

See SMITH_FAMILY_INTEGRATION.md for details.

Documentation

• Architecture - 4-layer architecture overview
• Migration Guide - Migrating from GeoSuite
• Smith Family Integration - PlotSmith, AnomSmith, TimeSmith integration
• GeoSuite Migration Assessment - What's been migrated from GeoSuite