Predicting vehicle prices via quantum-assisted feature selection

Abstract Feature selection is a technique used to reduce complexity and improve performance in regards generalization, fit, accuracy of prediction for machine learning models. A central challenge this process that the search over space features find subset k optimal known NP-Hard problem. In work, we study metrics encoding combinatorial as binary quadratic model, such Generalized Mean Information Coefficient Pearson Correlation application underlying regression problem price prediction. We compare predictive model leveraging quantum-assisted vs. classical subroutines search, using minimum redundancy maximal relevancy heuristic our approach. cross validate models on real world prediction, show improvement mean absolute error scores method $$(1471.02 \pm {135.6})$$ ( 1471.02 ± 135.6 ) similar methodologies greedy $$(1707.2 {168})$$ 1707.2 168 , recursive feature elimination $$(1678.3 {143.7})$$ 1678.3 143.7 all $$(1546.2 {154}$$ 1546.2 154 ). Our findings by quantum assisted routines solutions which increase quality output while reducing input dimensionality algorithm synthetic real-world data.