Machine Learning Tools to Time Series Forecasting
Karinne Ramírez-Amaro, J. C. Chimal-Eguía · 2007
In this paper a new input representation of the data of the time series and a new learning approachis presented. The input data representation is based on the information obtained by thedivision of image axis of the time series into boxes. Then, this new information is implemented ina new learning technique which through probabilistic mechanism this learning could be applied tothe interesting forecasting problem. The results indicate that using the methodology proposed inthis article it is possible to obtain forecasting results with good enough accuracy.