This study conducts nowcasting of monthly industrial activity deflators for the service and retail sales sectors using high-dimensional explanatory variables, and compares the performance of various linear and nonlinear forecasting methods. Linear approaches combine AR/ARX models with factor analysis and shrinkage estimation techniques, while nonlinear approaches include tree-based methods and neural network models. The results show that the AR + Ridge approach performs best for the service deflator, whereas the ARX + factor analysis with LASSO/Elastic Net yields the strongest performance for the retail sales deflator. In addition, incorporating a one-year lag improves forecasting accuracy for both deflators when newly released information is available. Overall, the findings highlight the importance of methodological structure and the timing of explanatory variables in nowcasting industrial activity deflators.