Regression

Classification predicts a category, regression predicts a continuous number, tomorrow's temperature, a house's price, a patient's blood sugar level. The core workflow, fit/predict/evaluate, stays the same, only the kind of output and the evaluation metrics change.

Predicting a number

from sklearn.datasets import load_diabetes
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error, r2_score

diabetes = load_diabetes()
X_train, X_test, y_train, y_test = train_test_split(
    diabetes.data, diabetes.target, test_size=0.2, random_state=42
)

model = LinearRegression()
model.fit(X_train, y_train)
predictions = model.predict(X_test)

print('Predicted:', predictions[:5].round(1))
print('Actual:   ', y_test[:5])

load_diabetes() is another small bundled dataset, this one maps patient measurements to a numeric disease progression score, a continuous value, not a category, so classification metrics like accuracy don't apply. LinearRegression learns a straight-line (technically, a weighted sum) relationship between the input features and that number.

Regression metrics

mse = mean_squared_error(y_test, predictions)
r2 = r2_score(y_test, predictions)
print('MSE:', mse)
print('R^2:', r2)

• MSE (mean squared error) averages the squared difference between each prediction and the true value, squaring both penalizes big misses more than small ones and keeps the result positive. Lower is better, but MSE alone is hard to interpret since its units are "target units squared".
• R² (R-squared) is easier to read: it's the fraction of the target's variance your model explains, 1.0 means perfect predictions, 0.0 means your model is no better than always guessing the average, and it can even go negative for a model worse than that baseline.