CompTIA DataAI Exam DY0-001

DataAI (V1) exam objectives summary

Mathematics and statistics (17%)

• Statistical methods: applying t-tests, chi-squared tests, analysis of variance (ANOVA), hypothesis testing, regression metrics, gini index, entropy, p-value, receiver operating characteristic/area under the curve (ROC/AUC), akaike information criterion/bayesian information criterion (AIC/BIC), and confusion matrix.
• Probability and modeling: explaining distributions, skewness, kurtosis, heteroskedasticity, probability density function (PDF), probability mass function (PMF), cumulative distribution function (CDF), missingness, oversampling, and stratification.
• Linear algebra and calculus: understanding rank, eigenvalues, matrix operations, distance metrics, partial derivatives, chain rule, and logarithms.
• Temporal models: comparing time series, survival analysis, and causal inference.

Modeling, analysis, and outcomes (24%)

• EDA methods: using exploratory data analysis (EDA) techniques like univariate and multivariate analysis, charts, graphs, and feature identification.
• Data issues: analyzing sparse data, non-linearity, seasonality, granularity, and outliers.
• Data enrichment: applying feature engineering, scaling, geocoding, and data transformation.
• Model iteration: conducting design, evaluation, selection, and validation.
• Results communication: creating visualizations, selecting data, avoiding deceptive charts, and ensuring accessibility.

Machine learning (24%)

• Foundational concepts: applying loss functions, bias-variance tradeoff, regularization, cross-validation, ensemble models, hyperparameter tuning, and data leakage.
• Supervised learning: applying linear regression, logistic regression, k-nearest neighbors (KNN), naive bayes, and association rules.
• Tree-based learning: applying decision trees, random forest, boosting, and bootstrap aggregation (bagging).
• Deep learning: explaining artificial neural networks (ANN), dropout, batch normalization, backpropagation, and deep-learning frameworks.
• Unsupervised learning: explaining clustering, dimensionality reduction, and singular value decomposition (SVD).

Operations and processes (22%)

• Business functions: explaining compliance, key performance indicators (KPIs), and requirements gathering.
• Data types: explaining generated, synthetic, and public data.
• Data ingestion: understanding pipelines, streaming, batching, and data lineage.
• Data wrangling: implementing cleaning, merging, imputation, and ground truth labeling.
• Data science life cycle: applying workflow models, version control, clean code, and unit tests.
• DevOps and MLOps: explaining continuous integration/continuous deployment (CI/CD), model deployment, container orchestration, and performance monitoring.
• Deployment environments: comparing containerization, cloud, hybrid, edge, and on-premises deployment.

Specialized applications of data science (13%)

• Optimization: comparing constrained and unconstrained optimization.
• NLP concepts: explaining natural language processing (NLP) techniques like tokenization, embeddings, term frequency-inverse document frequency (TF-IDF), topic modeling, and NLP applications.
• Computer vision: explaining optical character recognition (OCR), object detection, tracking, and data augmentation.
• Other applications: explaining graph analysis, reinforcement learning, fraud detection, anomaly detection, signal processing, and others.