What is Predictor?

Predictor: Advanced Machine Learning for Material Properties

Predictor is a machine learning system engineered to forecast material properties across a broad spectrum of materials including steels, non-ferrous alloys, and polymers. Utilizing extensive datasets from Total Materia Horizon curated database, the system employs proprietary methodologies for data curation, classification, and normalization to predict physical and mechanical properties with high precision.

Key Features:

• High Accuracy: Predictor achieves a relative error rate of 5% or less, ensuring precise results for engineering applications, such as CAE (Computer-Aided Engineering) and FEA (Finite Element Analysis), making it ideal for simulations and performance analysis.

• Data Gap Bridging: One of Predictor's standout features is its ability to fill in data gaps where experimental data may be missing. By utilizing machine learning, the system provides reliable estimates for material properties that would otherwise be unavailable, significantly expanding the range of usable data.

• Advanced ML Architectures: Incorporates cutting-edge machine learning technologies, such as XGBoost and TabNet, to manage large and complex datasets. These advanced architectures ensure high-quality predictions by efficiently handling data variability and improving model performance.

• Specialized Models: The development of Predictor involved creating multiple, tailored machine learning models for specific material families and properties. This specialization improves the accuracy and relevance of predictions for different applications, whether you're dealing with metals, polymers, or other material types.

Total Materia Predictor is designed with a user-friendly interface, enabling engineers to easily access and utilize predicted material properties in their analyses and simulations.