METHOD AND APPARATUS FOR ESTIMATING TREATMENT EFFECT BY USING PROPENSITY SCORE RESIDUAL TERM IN LEAST-SQUARE REGRESSION ANALYSIS

摘要

The present invention relates to a method for estimating a treatment effect by using a propensity score residual term in least-square regression analysis and an apparatus therefor, the method comprising the steps of: allowing a first residual term calculation unit to calculate a first parameter value and a propensity score on the basis of a treatment variable model for a treatment variable and a distribution model for a first error term included in the treatment variable model and calculate a first residual term by subtracting the propensity score from the treatment variable; allowing a second residual term calculation unit to calculate at least one second parameter value included in an approximation model for a conditional average value of a result variable on the basis of the first parameter value, generate a parameter-applied approximation model by applying the at least one second parameter value to the approximation model, and calculate a second residual term by subtracting the parameter-applied approximation model from the result variable; and allowing a treatment effect estimation unit to estimate, on the basis of the first residual term and the second residual term, a treatment effect parameter in which the treatment variable affects the result variable.