Incremental Sinusoidal Approximation of Time Series with LibreOffice Calc Solver

Financial time series forecasting has high importance in many decisionmaking situations. There are many methods and approaches for time series forecasting. Artificial Neural Networks are widely used. Support Vector Machines are giving even better results than artificial neural networks. Both tools are pretty complicated and have their advantages and disadvantages. This research proposes an approximation based on cumulative sinusoids. From the Fourier Transform, it is well known that each signal can be represented as a sum of sine functions. Optimal coefficients for the sine functions and the linear component, in this research, are searched with the Solver module of LibreOffice Calc. The search of the optimal values is done on an incremental basis. The result show that the time series can be successfully handled as signals.