Transfer Learning: Machine Learning Models and Clustering Configurations Performances in Time Series Forecasting

In the recent years Transfer Learning approaches have been widely implemented in several machine learning fields, such as Computer Vision, Natural Language Processing and Time Series Forecasting. The idea behind this technique is to improve the learning process of a specific task by using already acquired knowledge from another distinct one, in order to obtain better and faster results, requiring less domain specific resources. In this study will be observed different configurations of this method ,jointly with Global Machine Learning Models and Clustering algorithm, analyzing in which one of them it exploits better its potential in the task of Time Series Forecasting, experimenting with its application and trying to draw some significant conclusion about the nature, the results and the logic that lays behind its employment.