EN
Data processing algorithms are important parts of modern measurement systems. These algorithms are often delivered to the user as complex program and their numerical structure is not known. Therefore also influence of algorithm on processed data accuracy is not known. One of the methods to evaluate uncertainty propagation through algorithm is based on matrix form of algorithm. Coefficient matrix of algorithm represents its numerical operations and it is a basis to algorithm accuracy evaluation. The paper presents a method how to identify this coefficient matrix when algebraic form of the algorithm is known or is difficult to use. Identified matrix form of algorithm is then used to estimate uncertainty propagation through exemplary algorithms. Results are compared with experiments.