Crane Modeling

Next Generation Predictive Modeling Engine

Use automated machine learning and deep learning powered by CraneAI to discover new predictors, build deployable models, and conduct strategic analyses.

Build, iterate, and deploy models in lightning speed thanks to the optimizations of GeodatacubesTM and the ~limitless computing power of the cloud.
Fully automated for the non-specialist – but- heavily configurable for the expert.
Focus on model creation rather than experiment management thanks to our comprehensive integrated experiment tracking and all metadata capture.
Automatically incorporate 1000s of spatial relationships and statistics as explicit features in your model.

All model and research artifacts (model weights, definitions, training and testing results, etc.) are fully transparent and automatically archived.
Flexibility to export the models you create for use within your own pipeline or deploy them directly within the Paper Crane platform.
All machine learning models are delivered in python, giving you complete control and visibility over the models that you create.

Build optimal machine learning models using generic search across a validated library of model frameworks optimized for the large parameter space and data volumes typical in geospatial.
Fully automated ML model creation pipeline handles data split creation, hyperparameter optimization, and model performance testing.

Leverage integrated deep learning model library and imagery to perform identification, regression, and segmentation tasks.
Build new deep learning models using Paper Crane’s state-of-the art transfer-learning enabled CNN optimized for geospatial tasks.
Seamlessly couple machine learning and deep learning models.