hops-data-discovery

name	hops-data-discovery
description	Find available data in Hopsworks from feature groups, data sources, free-text search, and files.

Data Discovery on Hopsworks

Concept

Find available data in a Hopsworks project across four surfaces: feature groups (and views / training data), free-text search, data sources, and files. Prefer the hops CLI for data discovery over writing Python.

Discovery enables feature reuse: a model that reuses an existing feature does not need a new feature pipeline, so search the feature registry first before engineering features from scratch. The commands below browse that registry (definitions, schemas, tags, statistics).

Commands / API

Feature groups (and views / training data)

hops fg list                              # all feature groups in the project
hops fg info <name> --version 1           # metadata (online flag, PK, event time)
hops fg features <name> --version 1       # schema with PK / partition flags
hops fg preview <name> --version 1 --n 10 # first rows (note: preview <name>, not <name> preview)
hops fv list                              # feature views

Free-text search

hops search ls <term>                     # match FGs, FVs, training datasets, features
hops search ls <term> --type feature_group   # restrict by type
hops search ls <term> --global            # search across shared/other projects

Data sources (external connectors)

hops datasource list                      # configured connectors (Snowflake, S3, JDBC, ...)
hops datasource info <name>

Files (HopsFS)

hops files list [path]                    # browse the project filesystem

Related skills

• hops-trino-sql — run SQL against the offline tables you discover.
• hops-fg / hops-fv — create and read these via the Python SDK.
• hops-data-source — mount a new external source as a feature group.