SAS Viya Data Science and Model Deployment Training Course

Course overview

Organisations investing in SAS Viya need more than accurate models: they need repeatable ways to prepare governed data, compare modelling approaches, explain results to decision-makers, and deploy approved models into operational processes. Data scientists and analysts are often expected to bridge these stages while working across CAS tables, visual pipelines, code, model registries and production scoring environments. This course addresses that practical gap by showing participants how to move a business use case from raw data to a governed, deployable analytical asset in SAS Viya.

Participants work through the SAS Viya data science workflow using SAS Studio, Cloud Analytic Services (CAS), Model Studio and SAS Model Manager. They learn to load and profile data in CAS, engineer features, build supervised machine-learning pipelines, assess model performance, compare challenger models, and interpret model drivers. The course also covers project organisation, model registration, documentation, bias and performance considerations, publishing options, and post-deployment monitoring. Participants gain both point-and-click and code-based experience, enabling them to select the most appropriate Viya interface for the task.

Instructor-led demonstrations are followed by guided labs built around a realistic customer-response prediction case. Each participant develops a documented model pipeline, evaluates it against business success criteria, registers the selected model, and prepares a deployment and monitoring plan. The final deliverable is a model handover pack containing the model comparison rationale, scoring inputs and outputs, deployment route, validation evidence, and monitoring measures that can be adapted for workplace use.

The course is suited to professionals who already work with data and want to use SAS Viya as a practical environment for machine learning and governed model deployment. It is equally valuable for managers responsible for establishing consistent analytics delivery practices across data science, risk, marketing, operations or fraud teams.

Course objectives

By the end of this course, participants will be able to:

• Load, inspect and manage analytical data in CAS using libraries, caslibs and in-memory tables
• Prepare modelling data by treating missing values, transforming variables and creating derived features
• Build supervised learning pipelines in SAS Model Studio for classification and prediction use cases
• Write and run SAS code in SAS Studio to query, manipulate and score CAS data
• Evaluate candidate models using lift, ROC, confusion matrix, misclassification and variable-importance outputs
• Interpret model results with partial dependence, feature contribution and business performance measures
• Register approved models in SAS Model Manager with version, input, output and documentation metadata
• Produce a model deployment and monitoring plan covering publishing, validation, drift and performance thresholds

Benefits of attending

Target competencies

Who should attend

• Data Scientists — who need to build, assess and operationalise models within SAS Viya
• Data Analysts — who are moving from reporting and statistical analysis into predictive modelling
• Analytics Managers — who need a governed workflow for approving and deploying analytical models
• Risk Analysts — who develop scorecards or predictive models requiring traceability and monitoring
• Marketing Analysts — who use propensity, churn or next-best-action models to target customer activity
• Data Engineers — who support CAS data preparation and production model-scoring pipelines

Requirements and prerequisites

Participants should be comfortable working with tabular data and understand basic analytical concepts such as rows, columns, data types, missing values, target variables and train/test data splits. Familiarity with descriptive statistics and at least one of SQL, SAS programming, Python or R is helpful, particularly for the SAS Studio labs, but advanced programming is not required. Participants should have used spreadsheets, databases or BI tools to investigate data. No prior SAS Viya, SAS Model Studio, SAS Model Manager or machine-learning deployment experience is required. Complete beginners to data analysis should first gain basic data literacy and introductory statistics.

Training methodology

The course combines focused instructor demonstrations with individual SAS Viya labs in a configured training environment. Participants alternate between SAS Studio coding tasks and Model Studio visual pipelines, using CAS tables throughout the workflow. A running customer-response case provides the data and business context for profiling, feature engineering, model comparison and deployment decisions. Facilitated reviews require participants to justify their selected model using performance and governance evidence. On the final day, participants convert their work into a practical model handover and monitoring plan for their own organisational context.

Course outline

Day 1: SAS Viya data science foundations

• SAS Viya architecture and the role of Cloud Analytic Services
• Navigating SAS Environment Manager, SAS Drive and project assets
• CAS sessions, caslibs and in-memory table concepts
• Loading files and promoting tables for shared analytical use
• Exploring data with SAS Studio and CAS-enabled procedures
• Understanding analytical project roles, access and asset organisation
• Defining business objectives, target variables and model success criteria

Workshop: Participants load a customer-response dataset into CAS, profile its key fields, and produce a concise modelling problem statement with agreed success measures.

Day 2: Data preparation and feature engineering

• Data quality assessment for missing, invalid and duplicate values
• Partitioning data into training, validation and test samples
• Variable transformations and binning for analytical features
• Categorical encoding and handling high-cardinality variables
• Feature creation using SAS Studio DATA step and PROC SQL
• CAS data manipulation with FedSQL and CAS actions
• Building reusable data-preparation nodes in SAS Model Studio

Workshop: Participants create a prepared CAS modelling table, document transformation decisions, and build a repeatable data-preparation flow.

Day 3: Machine learning pipelines and model evaluation

• Model Studio pipeline structure, data nodes and modelling nodes
• Decision tree and forest models for classification
• Gradient boosting and neural network model options
• Logistic regression as an interpretable benchmark model
• Hyperparameter tuning and automated model selection
• ROC curves, lift charts and confusion-matrix interpretation
• Comparing champion and challenger models against business thresholds

Workshop: Participants build and compare multiple classification models in Model Studio, then nominate a champion model with a written selection rationale.

Day 4: Model governance and deployment preparation

• Model interpretability using variable importance and partial dependence
• Model bias, fairness and population-risk considerations
• Model documentation, assumptions and reproducibility evidence
• Registering models, versions and metadata in SAS Model Manager
• Creating model input and output definitions for scoring
• Publishing routes through SAS Micro Analytic Service and SAS Container Runtime
• Validation, approval and rollback controls before production release

Workshop: Participants register their champion model in Model Manager and assemble a deployment checklist covering scoring, approvals and validation.

Day 5: Monitoring and operational model handover

• Production scoring patterns for batch, real-time and container deployment
• Model performance monitoring and outcome-data collection
• Data drift, prediction drift and concept-drift indicators
• Setting monitoring thresholds, alerts and review frequencies
• Champion-challenger testing and model refresh decisions
• Communicating model outputs to business owners and operational teams
• Planning a SAS Viya model lifecycle for a workplace use case

Workshop: Participants complete a model handover pack containing their deployment route, monitoring metrics, ownership model and scheduled review plan.

Tools & standards covered

SAS Studio, SAS Cloud Analytic Services (CAS), SAS Model Studio, SAS Model Manager

A typical training day

Live online deliveries follow the same structure in the East Africa Time zone, with shorter screen blocks and longer breaks.

What the fee includes

How you can take this course

Certification

Participants who complete the full five days receive the Skillset Development Certificate of Completion, stating the course title, course code, dates and delivery format — suitable for professional-development records and employer reimbursement.

Frequently asked questions