A comparison between the two main components regression methods and the wavelet estimation method for the semi-parametric partial linear model

Abstract

Many researchers have increased their interest in semi-parametric methods due to the diversity of data for the study community, where semi-parametric models provide a link between parametric and non-parametric models, as well as achieve estimates of a high level. It is the waveform estimation method (WAVE) and the main components regression method (PCR), where we use the simulation method and different sample sizes (128, 256, 512) and different variances (0.5, 1, 2) to compare the two methods and choose which is better, and through the simulation results and the graph, the researcher concluded that the best method is the wavelet estimation method (WAVE), which is where it overcame the principal components method (PCR) when the number of variables is four variables and eight variables and when using linear and quadratic smoothing functions.