Loading demo links...
What to study for each DSP-210 domain
Defining the need to be addressed through the application of data science
Covers translating a business or organisational problem into a data science need, including stakeholder requirements, decision context, and constraints. It also addresses defining measurable objectives and success criteria and assessing feasibility before data work begins.
Key skills
problem framing, stakeholder requirements, success criteria, feasibility assessment, scope definition
Extracting, Transforming, and Loading Data
Covers extracting data from varied sources, transforming and cleaning it, and loading it into target systems with quality controls. It also addresses incremental loading, schema validation, deduplication, lineage, and handling errors and late-arriving data.
Key skills
ETL and ELT design, data quality validation, schema checks, deduplication, incremental loads, data lineage
Performing exploratory data analysis
Covers profiling and summarising data, examining distributions, relationships, missing values, and outliers, and using visualisation to reveal patterns. It also addresses grouped and time-based exploration that informs cleaning, feature, and modelling decisions.
Key skills
summary statistics, distributions, correlation, visualisation, missing data analysis, outlier detection
Building models
Covers building regression, classification, and clustering models with sound train and test discipline and effective feature engineering. It also addresses encoding, scaling, regularisation, handling imbalance, and controlling overfitting during model construction.
Key skills
regression, classification, clustering, feature engineering, encoding and scaling, regularisation, pipelines
Testing models
Covers testing built models with appropriate metrics, confusion matrix analysis, and cross-validation to estimate generalisation. It also addresses recognising overfitting and underfitting and selecting evaluation methods suited to imbalanced problems.
Key skills
classification and regression metrics, cross-validation, confusion matrix, overfitting detection, ROC-AUC
Operationalizing the pipeline
Covers deploying models and pipelines into production through batch or online serving and operating them reliably. It also addresses monitoring performance and drift, versioning, alerting on failures, and triggering retraining on evidence.
Key skills
model deployment, pipeline orchestration, monitoring and drift detection, versioning, retraining triggers
Communicating findings
Covers presenting findings, methods, and limitations clearly to stakeholders through appropriate visualisations and reports. It also addresses leading with recommendations, communicating uncertainty honestly, and avoiding misleading representations.
Key skills
reporting, visualisation choice, executive summaries, communicating uncertainty, stakeholder presentation