Skip to content

Regression vs Classification

Every supervised ML problem falls into one of two categories — the distinction is determined entirely by the type of your target variable.

Classification

Your target variable is categorical — it belongs to a discrete set of classes.

Example Target Classes
Spam detection is_spam 0 (No), 1 (Yes)
Disease diagnosis condition "Healthy", "Diabetic", "Pre-diabetic"
Customer segment tier "Gold", "Silver", "Bronze"

Algorithms: LogisticRegression, DecisionTreeClassifier, RandomForestClassifier, SVC

Metrics: Accuracy, Precision, Recall, F1-Score, ROC-AUC

Regression

Your target variable is continuous — it can take any numeric value within a range.

Example Target Range
House pricing sale_price £50,000 – £2,000,000
Temperature forecasting temp_celsius -30.0 – 50.0
Revenue prediction monthly_revenue £0 – £10,000,000

Algorithms: LinearRegression, Ridge, Lasso, RandomForestRegressor, GradientBoostingRegressor

Metrics: MAE, MSE, RMSE, R²

The Decision Rule

Ask yourself one question: "Can I meaningfully average two target values?"

  • If averaging makes sense (e.g., the average of £200k and £300k is £250k), it is regression.
  • If averaging is nonsensical (e.g., the average of "Cat" and "Dog" is meaningless), it is classification.

Common Pitfall

Binary targets encoded as 0/1 are still classification, not regression — even though they look numeric. Use LogisticRegression, not LinearRegression.

KSB Mapping

KSB Description How This Addresses It
K4.1 Statistical models and methods Understanding the statistical basis of regression and classification
K4.2 ML and AI techniques Implementing and comparing supervised learning algorithms
K4.4 Resource constraints and trade-offs Model complexity vs interpretability; computational cost
S1 Scientific methods and hypothesis testing Formulating hypotheses and testing with rigorous validation
S4 Building models and validating Cross-validation, train/test evaluation, performance metrics
B5 Impartial, hypothesis-driven approach Honest evaluation of model performance and limitations