Skip to learning content

DATA SCIENCE PYTHON PLAYGROUND

Machine Learning · Learn / Refresh

← Choose a deck
Machine Learning · Independent ML briefs

Workflow Challenges

Assemble a complete workflow from a realistic question and inspectable evidence.

Choose any brief. Each runs independently with its supplied inputs. Prerequisites and time estimates are guidance.

19 independent briefsOpen any case file
ML-X01
Regression

Explain a simple regression result

Model life expectancy from GDP per capita in the 2007 country snapshot. Report predictive evidence and the limits of interpreting the fitted line.

CSV Complete workflow and evidence
30–45 minutessimple_linear · validation · reference · final-test discipline
Open brief
ML-X02
Regression

Is a curve justified?

Compare a line and a polynomial workflow for Candy winpercent from sugarpercent. Decide whether the extra flexibility is supported.

CSV Complete workflow and evidence
40–50 minutessimple_linear · polynomial · validation · reference · final-test discipline
Open brief
ML-X03
Regression

Mixed predictors, two regression families

Compare multiple linear regression and a regression tree for Candy popularity using numeric and binary characteristics.

CSV Complete workflow and evidence
40–50 minutesmultiple_linear · regression_tree · validation · reference · final-test discipline
Open brief
ML-X04
Time-ordered prediction

Predict in the direction of time

Compare linear and tree workflows for later Seoul hourly bike demand. Use the declared weather and calendar scenario and state prediction-time availability assumptions.

CSV Complete workflow and evidence
45–60 minutesmultiple_linear · regression_tree · validation · reference · final-test discipline
Open brief
ML-X05
Classification

A classifier with a possible shortcut

Classify Penguin species from measurements and context. Compare a training-only measurements-only ablation to investigate reliance on geography/context.

CSV Complete workflow and evidence
30–45 minuteslogistic · validation · reference · final-test discipline
Open brief
ML-X06
Classification

Diagnose an overfitting classification tree

Build and diagnose a Car acceptability tree. Compare depth choices and explain class errors and the limits of feature importance.

CSV Complete workflow and evidence
30–45 minutesclassification_tree · validation · reference · final-test discipline
Open brief
ML-X07
Classification

Choose neighbours on a meaningful scale

Classify Penguin species with KNN using continuous measurements. Choose k without allowing held-away rows to become neighbours.

CSV Complete workflow and evidence
30–45 minutesknn_cls · validation · reference · final-test discipline
Open brief
ML-X08
Classification

Validate a nonlinear margin classifier

Use the five-feature Breast measurement scenario to validate a scaled RBF SVM and explain its available outputs.

CSV Complete workflow and evidence
30–45 minutessvm_cls · validation · reference · final-test discipline
Open brief
ML-X09
Classification

One-R for categorical decisions

Use the production One-R helper to classify Car acceptability from categorical inputs. Explain the selected rule and unknown-category fallback.

CSV Complete workflow and evidence
30–45 minutesone_r · validation · reference · final-test discipline
Open brief
ML-X10
Classification

One-R with numeric bins

Classify Candy popular using sugarpercent, pricepercent and the nine flags. Keep winpercent out of X and learn numeric bins only inside fits.

CSV Complete workflow and evidence
30–45 minutesone_r · validation · reference · final-test discipline
Open brief
ML-X11
Classification

Gaussian Naive Bayes for measurements

Build a Gaussian NB workflow for the five continuous Breast measurements. Interpret class probabilities and the independence approximation.

CSV Complete workflow and evidence
30–45 minutesnaive_bayes · validation · reference · final-test discipline
Open brief
ML-X12
Classification

Bernoulli Naive Bayes for binary characteristics

Classify Candy popular using only the nine presence/absence flags. Inspect smoothing and minority errors.

CSV Complete workflow and evidence
30–45 minutesnaive_bayes · validation · reference · final-test discipline
Open brief
ML-X13
Classification

Categorical Naive Bayes using the production path

Classify Car acceptability using the production categorical OHE–Bernoulli NB pipeline, preserving a reusable category schema.

CSV Complete workflow and evidence
30–45 minutesnaive_bayes · validation · reference · final-test discipline
Open brief
ML-X14
Classification

Shared and class-specific covariance

Compare LDA and regularised QDA on the same Penguin measurement population and stratified folds. Explain covariance assumptions and stability evidence.

CSV Complete workflow and evidence
40–50 minuteslda · qda · validation · reference · final-test discipline
Open brief
ML-X15
Neural networks

A neural regressor with trustworthy units

Predict Wine quality using the fixed 600-row teaching sample, feature scaling and an internal target transformation. Inspect convergence and original-unit errors.

CSV Complete workflow and evidence
45–60 minutesmlp_reg · validation · reference · final-test discipline
Open brief
ML-X16
Neural networks

Does a neural classifier justify its complexity?

Compare a small neural classifier with logistic regression on the five-feature Breast scenario. Consider predictive evidence, convergence and cost.

CSV Complete workflow and evidence
45–60 minutesmlp_cls · logistic · validation · reference · final-test discipline
Open brief
ML-X17
Clustering

Discover and describe Penguin groups

Discover groups from Penguin measurements with species hidden during fitting. Combine k evidence with sizes and original-unit profiles.

CSV Complete workflow and evidence
30–45 minuteskmeans · discovery · profiles
Open brief
ML-X18
Clustering

Describe a sampled hierarchy

Describe a Ward hierarchy of a reproducible sample of at most 500 Breast measurement rows, with diagnosis excluded from fitting.

CSV Complete workflow and evidence
30–45 minuteshierarchical · discovery · profiles
Open brief
ML-X19
PCA

Retain information without confusing it with a picture

Reduce all 30 Breast measurements with PCA. Retain at least 90% variance and distinguish the reduced representation from a separate two-dimensional view.

CSV Complete workflow and evidence
30–45 minutespca · discovery · profiles
Open brief