Classification vs. Regression Revision Notes

Question 1

Definition of Classification

Answer 1

Classification is a supervised learning approach that categorizes data into predefined classes or labels. It predicts discrete outcomes, typically binary (e.g., yes/no) or multiclass (e.g., types of fruits).

Question 2

Definition of Regression

Answer 2

Regression is a supervised learning method used to predict a continuous outcome (numeric value) based on one or more variables. It establishes the relationship between dependent and independent variables.

Question 3

What are the key differences between Classification and Regression?

Answer 3

Output Type: Classification predicts discrete labels, while Regression predicts continuous quantities.
Evaluation Metrics: Classification uses accuracy, confusion matrix, precision, etc., whereas Regression uses mean squared error, R-squared, etc.
Algorithms: Common Classification algorithms include Logistic Regression, Decision Trees, SVMs. Regression algorithms include Linear Regression, Polynomial Regression.

Question 4

What are the key similarities between Classification and Regression?

Answer 4

Both are types of supervised learning.
They require historical data for model training.
They aim to make predictions based on input features.
Both can use similar techniques for feature selection, cross-validation, and overfitting prevention.

Question 5

What are practical scenarios for applying Classification?

Answer 5

Email spam detection (binary: spam/not spam)
Image recognition (multiclass: categorizing images)
Loan default prediction (binary: default/no default)
Medical diagnosis (binary or multiclass: diagnosing diseases)

Question 6

What are practical scenarios for applying Regression?

Answer 6

Predicting house prices based on various features (continuous: price estimation)
Forecasting stock market prices (continuous: value prediction)
Estimating life expectancy (continuous: age prediction)
Predicting sales in retail (continuous: sales volume)