AI-ordlista
Den kompletta ordlistan över AI
Propensity Score Matching
Causal inference technique using propensity scores to create pairs of treated and untreated units with similar probabilities of receiving treatment.
Exact Matching
Matching method where treated and untreated units are matched on identical values of specific covariates.
Nearest Neighbor Matching
Matching technique that selects for each treated unit the closest control unit according to a specified distance metric.
Caliper Matching
Variant of nearest neighbor matching where only pairs with a distance below a predefined threshold (caliper) are accepted.
Optimal Matching
Matching method that minimizes the total overall distance between all matched pairs using combinatorial optimization algorithms.
Mahalanobis Matching
Technique using Mahalanobis distance to match units based on multivariate covariates, taking into account their correlations.
Kernel Matching
Weighted matching method where each treated unit is matched to multiple controls weighted by a kernel function based on distance.
Coarsened Exact Matching
Coarsening technique where continuous covariates are discretized into strata before exact matching on these categories.
Radius Matching
Matching method where all control units within a predefined radius around each treated unit are included and averaged.
Genetic Matching
Adaptive matching algorithm using genetic principles to optimize covariate balance between matched groups.
Matching Distance
Quantitative metric measuring the dissimilarity between two potentially matched units, fundamental for matching quality.
Full Matching
Optimal matching technique where units are grouped into subsets with variable treatment:control ratios to minimize distance.
Matching Diagnostics
Set of statistical tests and visualizations assessing the quality and adequacy of the matching performed on the data.