Machine Learning curriculum

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Define prediction target, label timing, features, unit of analysis, baseline, and decision use.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Turn a vague model idea into a supervised-learning spec with target, features, split rule, and baseline.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Detect leakage and ambiguous labels before modeling.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Fit mean, linear, and regularized regression baselines.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Compare a regression model to a baseline and explain whether the improvement is meaningful.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Use MAE, RMSE, residuals, and calibration-style checks.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Use logistic regression, confusion matrices, precision, recall, specificity, ROC-AUC, and threshold tradeoffs.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Choose a classification threshold from a confusion matrix and defend the tradeoff.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Choose metrics based on false-positive and false-negative costs.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Use decision trees, random forests, gradient boosting concepts, feature importance, and pruning.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Explain how a tree split improves prediction and how validation prevents overfitting.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Recognize when tree models overfit.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Use clustering, dimensionality reduction, embeddings, and similarity measures.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Interpret a clustering result with one useful insight and one reason it may be misleading.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Validate unsupervised patterns without pretending they are ground truth.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Use layers, activation, loss, gradient descent, backpropagation intuition, and regularization.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Sketch a small neural network for tabular or image input and explain loss, output, and overfitting controls.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Understand what neural networks buy and what they make harder.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Use train/validation/test splits, cross-validation, leakage audits, drift checks, and fairness metrics.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Design a validation protocol that blocks leakage and reports at least one subgroup performance check.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Separate model selection from final evaluation.

Start with Principles of Data Science and 6.0002 Introduction to Computational Thinking and Data Science. Take notes until you can explain: Build a small end-to-end ML project with reproducible data prep, model comparison, evaluation, and documentation.

Do the first sequenced checks until the definitions, vocabulary, and setup are correct without hints.

Produce the assignment artifact, then pass the application and analysis checks tied to: Write a model card with intended use, data limits, metrics, fairness checks, and monitoring plan.

Any miss becomes an error-log entry, a clean redo, and one nearby transfer problem before advancing.

Bring your attempted work, the exact missed check, and one question about: Plan monitoring for data drift, error shifts, and retraining triggers.