Machine Learning Parameters¶
We present in what follows the parameters which apply specifically to the Machine Learning model type.
Subtype¶
We support the following Subtypes:
- Regression 1 (default): predict a property (e.g. Band gap) of the target material given a relationship between material features (descriptors) of similar materials (training set), and same property of the materials in the training set.
- Classification 2: predict which category (e.g. superconductor, etc) a target material belongs to based on the descriptors of similar materials in the training set.
- Clustering 3: given a set of descriptors, partition the data into several sets, such that materials within one set are more similar to one-another than they are to materials in another set.