In this talk, we will discuss additive partial linear models with symmetric autoregressive errors, proposed for the modeling of time series. A main feature of this class of models is its capacity to handle explanatory variables with both linear and nonlinear structures. Furthermore, the inclusion of conditional symmetric errors enables the modeling of data exhibiting high-order correlation, as well as error distributions with heavier or lighter tails compared to the normal distribution. The details of the model and its estimation are examined, along with the limitations in forecasting and proposals for predicting future values. Finally, several applications of this approach will be presented. The presentation will delve into the details of the model and its estimation procedures, while also addressing the limitations associated with forecasting and proposing strategies for predicting future values.