Skip to content

Support real-valued features #22

@fritzo

Description

@fritzo

Support real-valued features for latent categorical and latent Gaussian states. For latent categorical states, this will correspond to a mixture of Gaussians under a normal-inverse-chi-squared prior. For latent Gaussian state, this will be a structured linear model.

  • PR Implement DataTable class for heterogeneous row-oriented data.
  • Implement ingest logic in format module.
  • Refactor training module to use DataTable
  • Refactor serving module to use DataTable
  • Refactor everything else to use DataTable

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions