MACHINE LE

Apuntes

MACHINE LEARNING BASIC CONCEPTS SUPERVISED MACHINE LEARNING CLASSIFICATION TASK. Algorithm Repaid loan = yes Years Amount Salary Own house? Overdue accounts? Repaid loan 15 60000 1900 Yes 2 No 2 30000 3500 Yes 0 Yes 9 9000 1700 Yes 1 No 15 18000 3000 No 0 Yes 10 24000 2100 No 0 No ... ... ... ... ... Years Amount Salary Own house? Overdue accounts? Repay loan 10 50000 3000 Yes 0 ?? IF OA >0 THEN NO IF OA==0 AND S>2500 THEN Yes T = training set Model future data Instances, examples, data points Attributes, features, predictors, Input variables, Independent variables, explanatory variables Label, class, output variable, dependent variable, response, predictand, target DEFINITIONS: ATTRIBUTES AND RESPONSE • Attributes / features / input variables • A feature is an individual measurable property of an instance • Types: • Numeric: (real numbers or integers): • 0, 1, 2 • 1.3, 7.9, 10.798, … • Categorical: yes / no ; red / green / yellow • Many ML algorithms cannot deal with them: they must be encoded into numbers (one-hot-encoding, ...) • Ordinal: cold, lukewarm, hot • Most ML cannot deal with them, they are encoded as integer numbers • Response / output variable • If categorical (e.g. c...