A collection of time series that are statistically independent of one another is called a multiple time series. Furthermore, missing data typically affects time series, meaning that some time series sample values are unknown at any particular time. The Iterated Transformer Encoder-based Time Series Prediction algorithm is presented in this article. It predicts multiple time series with missing data and simultaneously provides an uncertainty measure of the prediction. The algorithm uses iteratively the Correlation Dimension estimation using Grassberger-Procaccia algorithm for fixing the time series model order (i.e., the number of past samples needed to accurately model the time series), and the proposed Transformer Encoder-based Time Series Prediction to approximate the multiple time series skeleton. The proposed algorithm has been experimentally validated on a multiple time series with missing data that expresses the concentration of nitrogen dioxide in two different sites. The results show that the algorithm performs better than state-of-the-art competitors in terms of average mean squared error.