This study presents a machine learning (ML) model that accurately predicts the antigenic properties of influenza A virus (IAV) H3N2 isolates using their hemagglutinin subunit I (HAI) sequences and associated metadata. The model, trained on past seasons' data, learns an updated nonlinear mapping from genetic to antigenic changes, distinguishing antigenic variants and characterizing seasonal dynamics of HAI sites with the strongest influence on antigenic change. The model's predictions aid in influenza surveillance, public health management, and vaccine strain selection. It achieves a mean absolute error (MAE) of 0.702 antigenic units per season and robustly performs under data-limited scenarios. The model identifies key HAI sites, mostly within known epitopes, that contribute significantly to antigenic evolution, providing insights into the seasonal dynamics of these sites. The approach complements existing antigenic characterization efforts, enabling comprehensive global influenza antigenicity monitoring and improved vaccine strain selection.This study presents a machine learning (ML) model that accurately predicts the antigenic properties of influenza A virus (IAV) H3N2 isolates using their hemagglutinin subunit I (HAI) sequences and associated metadata. The model, trained on past seasons' data, learns an updated nonlinear mapping from genetic to antigenic changes, distinguishing antigenic variants and characterizing seasonal dynamics of HAI sites with the strongest influence on antigenic change. The model's predictions aid in influenza surveillance, public health management, and vaccine strain selection. It achieves a mean absolute error (MAE) of 0.702 antigenic units per season and robustly performs under data-limited scenarios. The model identifies key HAI sites, mostly within known epitopes, that contribute significantly to antigenic evolution, providing insights into the seasonal dynamics of these sites. The approach complements existing antigenic characterization efforts, enabling comprehensive global influenza antigenicity monitoring and improved vaccine strain selection.