# Time Series Prediction/Analysis

Time series analysis deserves a separate section, since results obtained in this field can be virtually applied to all problems characterized by continuously updating input data.

Time series forecasting is a challenging problem, that has a wide variety of application domains such as in engineering, environment, finance and others. When confronted with a time series forecasting application, typically a number of different forecasting models are tested and the best one is considered. Alternatively, instead of choosing the single best method, a wiser action could be to choose a group of the best models and then to combine their forecasts. In [9] the authors use a Multi-layer perceptron (MLP), Gaussian Processes Regression (GPR) and a Negative Correlation Learning (NCL) model. The authors of [8], instead, investigate the performance of using forecast combination in handling breaks in data series, observing how the performance of prediction strategies varies in presence of discontinuities and “holes” in time series. The problem of missing data is also dealt with in [94]. The authors propose a semi-supervised co-training method. Time series data are transformed to set of labeled and unlabeled data. Different predictors are used to predict the unlabeled data and the most confident labeled patterns are used to retrain the predictors further to and enhance the overall prediction accuracy. By labeling the unknown patterns the missing data is compensated for.