Resources

Titanic survival — features: Sex, Age, FirstClass, Children, Survived. Binary classification for ML-3.

Abalone: predict age (Rings) from physical measurements (Length, Diameter, Height, Weight). Regression.

Breast Cancer Wisconsin (UCI). Radiologic features + diagnosis (M/B → 0/1). Used in ML-3.

House prices (regression on area, number of rooms, etc.). Used in ML-2.

Atmospheric CO2 concentration over time. Regression / time series.

Synthetic cosine data to illustrate non-linear / polynomial regression.

The classic Iris (sepal/petal length/width → species). Beginner classification.

Monthly airline passengers, classic time series for ML-1.

Palmer Penguins (3 species, bill/wing measurements). Modern alternative to Iris.

UCI Mushroom (8,124 entries). Binary classification edible / poisonous from morphological descriptors only.

UCI Student Performance (math/Portuguese grades, family context). Regression or multiclass classification.

Credit card fraud detection (heavily imbalanced, ~0.17% fraud). Classic benchmark for minority-class problems.

Full Titanic (Kaggle), with more features than titanic_mini: Name, Ticket, Cabin, Fare, Embarked, …

Full Wisconsin Breast Cancer (30 features). Wider than cancer_mini.

Telco churn: predict customer cancellation from profile and usage data.

Star classification (temperature, luminosity, radius → spectral type). Multiclass, pedagogical.

UCI Adult / Census Income. Predict whether income > $50K/year (binary classification with bias to analyze).

Kaggle Ames Housing — complex regression on ~80 features, house prices in Ames (Iowa). For ML-5.

Kaggle Mercedes-Benz Greener Manufacturing: predict test duration on vehicles. Industrial regression.

Stroke prediction from medical and demographic variables. Imbalanced binary classification.

MNIST in CSV (70,000 28×28 handwritten digits, label 0-9). Multiclass classification, prefer local execution.

Sign Language MNIST: sign language alphabet as 28×28 pixel images.