IMPROVING THE RELIABILITY OF MACHINE LEARNING MODELS BY FILLING IN MISSING NAN VALUES IN MEDICAL DATASETS USING A GENETIC ALGORITHM

This article proposes a genetic algorithm-based approach to optimize the filling of missing NaN values in a dataset. The focus is on selecting NaN values in the dataset directly corresponding to the results of the classification task. In the proposed method, each individual is represented as a chromosome in the form of a vector of all missing values. The search space is bounded by the given intervals for numerical attributes, and by the set of appropriate categories for categorical attributes. The accuracy indicator of the Random Forest ensemble model was used as the fitness function in the genetic algorithm.