DataRobot Automated Machine Learning Training Course

Course overview

DataRobot enables teams to build and compare machine learning models without hand-coding every algorithm, but reliable use requires more than selecting “Autopilot.” Analysts and data professionals need to frame a business prediction problem correctly, prepare data that will not mislead a model, select suitable validation settings, and explain model results to decision-makers. This course addresses the practical gap between uploading a dataset and deploying a governed, useful predictive model for applications such as churn prevention, demand forecasting, credit risk, fraud detection, and service prioritisation.

Participants work through the DataRobot AI Platform workflow from data ingestion to model evaluation, prediction, and monitoring. They learn how to configure Autopilot; distinguish regression, classification, and time-aware modelling projects; inspect feature quality; interpret Leaderboard results; compare accuracy, calibration, and business-relevant metrics; use feature impact and prediction explanations; and assess fairness, data leakage, and drift risks. The course also covers model deployment patterns, batch scoring, prediction APIs, and the core controls provided through DataRobot MLOps.

Teaching combines instructor-led demonstrations with guided platform labs using realistic business datasets. Each participant builds a documented DataRobot project, evaluates candidate models against defined acceptance criteria, produces explainability evidence, and prepares an operational handover plan. The final deliverable is a model decision pack containing the project configuration, selected model rationale, validation results, prediction workflow, monitoring measures, and recommendations for deployment or further testing.

The course is suited to professionals who need to deliver or oversee machine learning use cases using DataRobot rather than develop models entirely through Python or R. It gives technical practitioners a repeatable modelling workflow and gives managers and product owners the evidence needed to challenge, approve, and govern automated model decisions.

Course objectives

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

• Configure a DataRobot Autopilot project for regression, binary classification, multiclass classification, or time-aware prediction.
• Prepare a modelling dataset by identifying target leakage, missing-value issues, incorrect feature roles, and unsuitable partitioning fields.
• Select and justify validation, holdout, and cross-validation settings for a business prediction use case.
• Interpret the DataRobot Leaderboard using AUC, LogLoss, RMSE, MAE, F1, calibration, and business cost considerations.
• Compare candidate models with feature impact, SHAP-based prediction explanations, ROC curves, lift charts, and residual analysis.
• Create a deployment-ready prediction workflow using DataRobot batch predictions and prediction API concepts.
• Define monitoring thresholds for data drift, prediction drift, model performance, and prediction-service health in DataRobot MLOps.
• Produce a model decision pack documenting data assumptions, selected model evidence, risk controls, and deployment recommendations.

Benefits of attending

Target competencies

Who should attend

• Data Analysts — who need to turn operational datasets into defensible predictive models using DataRobot.
• Data Scientists — who need a faster, governed workflow for testing and operationalising many model candidates.
• Business Intelligence Analysts — who support forecasting, segmentation, retention, or risk decisions with data-driven evidence.
• Machine Learning Engineers — who must understand DataRobot model deployment, scoring, and monitoring workflows.
• Analytics Managers — who approve modelling approaches and need to evaluate model quality, explainability, and delivery risk.
• Product Owners — who define AI-enabled products and must translate business outcomes into measurable prediction problems.

Requirements and prerequisites

Participants should be comfortable working with structured tabular data in CSV or database-export form and should understand basic spreadsheet concepts such as columns, records, filters, and data types. Familiarity with descriptive statistics, including averages, distributions, and percentages, is helpful. Participants should also be able to describe a business outcome that could be predicted from historical data. Prior Python, R, SQL, or machine learning programming experience is not required; the course uses DataRobot’s interface and guided exercises. Complete beginners should expect an intensive introduction to modelling terminology, evaluation metrics, and responsible use alongside platform practice.

Training methodology

Each day combines focused instructor explanation with live navigation of the DataRobot AI Platform and supervised hands-on labs. Participants work with prepared tabular business datasets to configure Autopilot, review model outputs, diagnose data and validation issues, and compare candidates on the Leaderboard. Short case discussions require participants to connect model metrics to business decisions such as retention targeting or demand planning. Small-group review sessions challenge model assumptions and governance choices. The final day uses an application-planning workshop to turn each participant’s model evidence into a practical deployment and monitoring proposal.

Course outline

Day 1: Framing predictive problems in DataRobot

• DataRobot AI Platform navigation and project workspace structure
• Business outcome definition and prediction target design
• Regression, binary classification, multiclass classification, and time-aware project types
• Training data, scoring data, target columns, and feature roles
• Data ingestion from CSV files and DataRobot AI Catalog concepts
• Data quality profiling, missingness review, and data type correction
• Target leakage identification and prevention

Workshop: Participants create a churn-prediction project, define the target and feature roles, and produce a data-readiness checklist.

Day 2: Building and validating AutoML models

• Autopilot modes, model blueprints, and experiment configuration
• Partitioning strategies for training, validation, and holdout data
• Cross-validation and validation settings for imbalanced datasets
• DataRobot Leaderboard structure and model comparison workflow
• Classification metrics including AUC, LogLoss, F1, precision, and recall
• Regression metrics including RMSE, MAE, and R-squared
• Prediction thresholds, confusion matrices, and business cost trade-offs

Workshop: Participants run Autopilot on a customer-retention dataset and select three candidate models using a documented metric scorecard.

Day 3: Explaining model behaviour and model risk

• Feature impact ranking and feature effect interpretation
• SHAP-based prediction explanations for individual predictions
• Partial dependence and feature effect plots
• ROC curves, precision-recall curves, lift charts, and gains charts
• Residual analysis and error pattern investigation
• Bias and fairness assessment using sensitive-feature considerations
• Data leakage, proxy variables, and model documentation controls

Workshop: Participants investigate a selected model’s drivers and errors, then produce an explainability and risk briefing for a non-technical sponsor.

Day 4: Deploying predictions and operating models

• Champion model selection and deployment approval criteria
• DataRobot deployment creation and prediction-server concepts
• Batch prediction workflows and scored-data outputs
• Real-time prediction API concepts and request-response design
• Prediction data collection and feedback-data requirements
• DataRobot MLOps service health and prediction monitoring
• Data drift, prediction drift, and model performance monitoring

Workshop: Participants design a batch-scoring deployment for weekly demand forecasts and create a monitoring plan with named alert thresholds.

Day 5: Governed DataRobot delivery planning

• Model decision records and reproducible project documentation
• Model governance roles, approval gates, and ownership assignment
• Retraining triggers and model replacement decisions
• Human review controls for high-impact predictions
• Communicating uncertainty, limitations, and acceptable use
• DataRobot project handover and stakeholder reporting
• Use-case prioritisation by value, feasibility, and operational readiness

Workshop: Participants assemble and present a model decision pack containing their selected model rationale, deployment design, monitoring controls, and next-step recommendations.

Tools & standards covered

DataRobot AI Platform, DataRobot AutoML, DataRobot MLOps, DataRobot Notebooks

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