IBM C1000-120 key terms

A containerized platform that bundles data governance, cataloging, transformation, and machine learning services on Red Hat OpenShift, letting teams run an integrated data and AI stack on private or public infrastructure.

A development environment inside Cloud Pak for Data where users create projects, build notebooks, train models, and deploy them with tools such as Jupyter, Spark, and AutoAI.

A governance and catalog service that inventories data assets, assigns business terms and classifications, enforces policies, and gives users a searchable self-service view of trusted data.

The model lifecycle service in Cloud Pak for Data that stores trained models, manages deployment spaces, serves online and batch scoring, and monitors deployed models for drift and fairness.

The shared identity and access layer for Cloud Pak for Data services, providing single sign-on, user and group management, and token-based authentication across the platform's components.

A collaborative workspace in Watson Studio that holds data assets, notebooks, models, and collaborators, and serves as the container in which analysis and model building happen.

A governed collection of data assets and metadata that a defined group of users can search, preview, and request access to, with membership and visibility controlled by administrators.

A cataloged representation of a data source, such as a table, file, or connection, together with its metadata, profile, quality information, and governance assignments.

A stored definition of how to reach an external data source, including host, port, credentials, and driver properties, which projects and catalog assets reuse instead of embedding credentials individually.

A controlled vocabulary of business terms with definitions, stewards, and relationships, used to give shared meaning to data and to link technical assets to agreed business concepts.

A reusable governance object such as a business term, data class, reference data set, policy, or rule that can be published to a catalog and enforced or attached to assets.

A named pattern or set of criteria used to identify sensitive data, such as credit card numbers or national identifiers, so that matching columns can be masked or restricted automatically.

A governance rule that pairs a data class with an action, such as denying access or masking values, so that assets containing matching data are governed consistently.

The transformation of sensitive values so users see substituted or redacted content while the underlying data remains usable for analysis, applied through governance rules or masking flows.

A recorded trace of where an asset came from and how it was transformed, showing upstream sources and downstream consumers so users can assess origin, impact, and trustworthiness.

The process of harvesting technical metadata, such as schemas, tables, and columns, from a connected source into a catalog so assets can be discovered and governed.

A check applied to a column or table, such as uniqueness or range validation, whose results are scored and reported so stewards can track whether data meets expectations.

Automated analysis of a data set that reports value distributions, null counts, distinct counts, and inferred types, helping users understand content and detect anomalies before use.

A collaborative environment in Watson Machine Learning that groups deployed models and their infrastructure, with members and roles controlling who can create, promote, or score models.

A hosted scoring endpoint that answers synchronous prediction requests, typically used by applications that need immediate responses, and that can be scaled or updated independently.

A scheduled or on-demand scoring job that processes a large data set and writes predictions to an output destination, suited to high-volume work where latency is not critical.

An automated modeling capability that explores data transformations, algorithms, and hyperparameter settings, ranks candidate pipelines, and produces a deployable model with minimal manual coding.

The systematic search over model configuration settings to find the combination that best improves a chosen evaluation metric during training.

The assessment of a trained model against held-out data using metrics such as accuracy, precision, recall, or area under the curve, used to compare candidates and decide promotion.

Ongoing measurement of a deployed model's inputs, outputs, accuracy, and fairness, alerting owners when drift, bias, or degradation crosses configured thresholds.

A decline in a deployed model's relevance caused by changes in input data or the relationship between inputs and outcomes, detected by comparing live scoring behavior to a baseline.

An authorization model where permissions are granted to roles, and users receive access by being assigned to those roles within a platform, project, or deployment space.

A per-workspace assignment such as viewer, editor, or admin that determines what a user can do inside a specific project or catalog, independent of platform-wide roles.

A generated credential tied to a user identity that lets scripts and command-line tools authenticate to platform APIs without storing an interactive password.

A command-line interface that wraps platform REST APIs, enabling users to create projects, manage deployments, and script administrative tasks from a terminal.

The HTTP-based programming interface exposed by platform services, used to automate operations such as asset creation, model deployment, and governance actions from external applications.

Per-user stored secrets, such as database passwords, that a connection uses on the user's behalf so shared connection definitions do not expose individual login details.

An interactive document mixing code, output, and narrative that runs in Watson Studio with selectable runtimes, commonly used for data preparation, exploration, and model development.

The software environment, including language version and installed libraries, in which a notebook or job executes, chosen per asset to match dependency requirements.

A scheduled or manually triggered execution of a notebook, script, or pipeline on platform compute, used to automate recurring data preparation and model training work.

A visual transformation tool that lets users shape and cleanse data through ordered steps, producing reusable output that can be saved and consumed by other assets.

A provisioned running copy of a platform service, such as a catalog or machine learning service, that users associate with projects and spaces to gain its capabilities.