Prediction of time series is one of the topics that receive significant interest because of its importance in various fields, especially when studying natural phenomena. In this research, the transformation function model was reconciled where it aims to use the genetic algorithm to estimate the parameters of the final transformation function model.  Also, it was used to predict future values for the time series of monthly averages of temperatures in Nineveh Governorate for the period (1985-2000) as an output series and wind speed as an input series. In Nineveh Governorate, they are not stable in average and variance; when taking the square root of the data and taking the first seasonal difference as well as the first normal difference, stability was achieved, and then showed a model of the transformation function as shown in the equation (17). This research showed that the model's final parameters were estimated using the genetic algorithm based on the standard error squares average. The best estimate was chosen for the parameters that correspond to the lowest value of the average error squares, and by using this model, monthly temperature rates were predicted. Predictive values were shown to be consistent with the original values of the series. By depending on the transformation function model shown in the above equation, monthly averages of the temperature were predicted for the next four months, and the prediction results were consistent with the original time series values, which indicates the efficiency of the model.